Issue 17540: Need to support infinite and indefinite time constructs (date-time-ftf) Source: EDM Council (Mr. Mike Bennett, mbennett(at)edmcouncil.org) Nature: Enhancement Severity: Significant Summary: The Date Time Vocabulary needs to be able to support statements about periods of time which extend indefinitely into the future, and also describe periods of time which will have begun at indeterminate times in the past. As an example of the former, it is possible and meaningful for a contract to make statements about commitments or rights which extend in perpetuity, such as “Perpetual Bonds” which are bonds that pay interest forever. In general, it is necessary to be able to make meaningful statements which embody the concept of “Forever”. Similarly, it is necessary to be able to make meaningful statements which have been going on in perpetuity up to the present time. The DTV specification does currently allow for making statements about infinite time going forward, but not about time periods which have started at some indefinite time in the past. Meanwhile there are three other issues in play which touch on this same area. These may have an impact on the current ability to make statements about infinite time into the future as well, depending on their final resolution. The issues which have a bearing or potential bearing on this matter are: - Issue 16992: “Corollaries to Axiom D.4 in 8.2.3 are misstated”; - Issue 16993: “no syntax for indefinite time periods (date-time-ftf)”; - Issue 16997: “forever is misdefined” In summary, the requirement that needs to be met with the resolution of this and the above-referenced issues is: - Extend ‘forever’ with two new concepts: 1. Indeterminate time in the past 2. Indefinite time in the future - Ensure that the concept “forever” can be adequately defined (per 16997) including with reference to the time axis; - That there is syntax for the specification of indeterminate time periods that began at some point in the past and last up until the present (per 16993) - That the restatement of the axiom and corollaries referenced in 16992 take account of the two concepts above (indeterminate time in the past and indefinite time in the future) Resolution: Revised Text: Actions taken: August 6, 2012: received issue Discussion: End of Annotations:===== m: webmaster@omg.org Date: 06 Aug 2012 14:00:44 -0400 To: Subject: Issue/Bug Report ******************************************************************************* Name: Mike Bennett Employer: EDM Council mailFrom: mbennett@edmcouncil.org Terms_Agreement: I agree Specification: Date Time Vocabulary (DTV) Section: various FormalNumber: dtc/12-xx-xx Version: Beta 2 Doc_Year: 2012 Doc_Month: August Doc_Day: Day Page: various Title: Need to support infinite and indefinite time constructs Nature: Enhancement Severity: Significant CODE: 3TMw8 B1: Report Issue Description: The Date Time Vocabulary needs to be able to support statements about periods of time which extend indefinitely into the future, and also describe periods of time which will have begun at indeterminate times in the past. As an example of the former, it is possible and meaningful for a contract to make statements about commitments or rights which extend in perpetuity, such as .Perpetual Bonds. which are bonds that pay interest forever. In general, it is necessary to be able to make meaningful statements which embody the concept of .Forever.. Similarly, it is necessary to be able to make meaningful statements which have been going on in perpetuity up to the present time. The DTV specification does currently allow for making statements about infinite time going forward, but not about time periods which have started at some indefinite time in the past. Meanwhile there are three other issues in play which touch on this same area. These may have an impact on the current ability to make statements about infinite time into the future as well, depending on their final resolution. The issues which have a bearing or potential bearing on this matter are: - Issue 16992: .Corollaries to Axiom D.4 in 8.2.3 are misstated.; - Issue 16993: .no syntax for indefinite time periods (date-time-ftf).; - Issue 16997: .forever is misdefined. In summary, the requirement that needs to be met with the resolution of this and the above-referenced issues is: - Extend .forever. with two new concepts: 1. Indeterminate time in the past 2. Indefinite time in the future - Ensure that the concept .forever. can be adequately defined (per 16997) including with reference to the time axis; - That there is syntax for the specification of indeterminate time periods that began at some point in the past and last up until the present (per 16993) - That the restatement of the axiom and corollaries referenced in 16992 take account of the two concepts above (indeterminate time in the past and indefinite time in the future) To: date-time-ftf@omg.org Subject: Date-Time Issues 17540, 16992, 16993, 16997 - indefinite time intervals X-KeepSent: F730BBD9:97319387-85257A71:0012AAA4; type=4; name=$KeepSent X-Mailer: Lotus Notes Release 8.5.3 September 15, 2011 From: Mark H Linehan Date: Wed, 5 Sep 2012 23:30:47 -0400 X-MIMETrack: Serialize by Router on D01ML604/01/M/IBM(Release 8.5.3FP2IF1|July 25, 2012) at 09/05/2012 23:30:49 X-Content-Scanned: Fidelis XPS MAILER x-cbid: 12090603-7182-0000-0000-00000282D100 This set of four proposed resolutions addresses the issues around "indefinite" time intervals. They are based on today's discussion between Mike and I. 17540 adds concepts 'primordial' and 'perpetuity' and renames & redefines 'forever' as 'eternity'. (I am not crazy about these terms, particularly 'primordial', which is an adjective.) 16993 adds verb concepts such as 'time interval through occurrence'. 16997 is a duplicate of 17540. 16992 is mostly about durations, but was blocked by 17540. This version rewrites the Notes proposed to be added to 'time interval'. During conversation, we thought the last two Corollaries need some change, but on further review I think they are ok as proposed. ----------------------------- Mark H. Linehan STSM, IBM Research Date-Time Issue 16992- Corollaries to Axiom D.4 in 8.2.3 are misstated.doc Date-Time Issue 16993 - no syntax for indefinite time periods.doc Date-Time Issue 16997 - forever is misdefined.doc Date-Time Issue 17540 - Need to Support Infinite and Indefinite Time Constructs.doc Disposition: Resolved OMG Issue No: 17540 Title: Need to support infinite and indefinite time constructs Source: Mike Bennett, EDM Council, mbennett@edmcouncil.org Summary: The Date Time Vocabulary needs to be able to support statements about periods of time which extend indefinitely into the future, and also describe periods of time which will have begun at indeterminate times in the past. As an example of the former, it is possible and meaningful for a contract to make statements about commitments or rights which extend in perpetuity, such as Perpetual Bonds which are bonds that pay interest forever. In general, it is necessary to be able to make meaningful statements which embody the concept of Forever. Similarly, it is necessary to be able to make meaningful statements which have been going on in perpetuity up to the present time. The DTV specification does currently allow for making statements about infinite time going forward, but not about time periods which have started at some indefinite time in the past. Meanwhile there are three other issues in play which touch on this same area. These may have an impact on the current ability to make statements about infinite time into the future as well, depending on their final resolution. The issues which have a bearing or potential bearing on this matter are: - Issue 16992: Corollaries to Axiom D.4 in 8.2.3 are misstated; - Issue 16993: no syntax for indefinite time periods (date-time-ftf); - Issue 16997: forever is misdefined In summary, the requirement that needs to be met with the resolution of this and the above-referenced issues is: 1. Extend forever with two new concepts: a. Indeterminate time in the past b. Indefinite time in the future 2. Ensure that the concept forever can be adequately defined (per 16997) including with reference to the time axis; 3. That there is syntax for the specification of indeterminate time periods that began at some point in the past and last up until the present (per 16993) 4. That the restatement of the axiom and corollaries referenced in 16992 take account of the two concepts above (indeterminate time in the past and indefinite time in the future) Resolution: The FTF decided that there is no value in distinguishing .indefinite. from .infinite.. It chose to add new concepts that provide the basis for time intervals that extend indefinitely into the past or the future. New concepts .primordial. and .perpetuity. are defined respectively as the earliest time interval and the latest time interval. .Eternity. (synonym .forever.) is defined as .primordial through perpetuity.. This permits formulations such as .primordial through today. and .2012 through perpetuity.. A tool might support these formulations with a syntax such as .. until today .. and .. from 2012 on ... Issue 16993 adds verb concepts such as .time interval until situation model.. .primordial. and .perpetuity. can substitute for the .time interval. role to enable formulations such as .primordial until the Industrial Revolution.. Revised Text: (These update instructions apply to the beta-2 specification). Add a new clause 8.1.8 at the end of clause 8.1. The new clause adds a figure, so the following figures of clause 8 must be renumbered. 8.1.8 Indefinite Time Intervals Indefinite time intervals provide the basis for describing time intervals that extend indefinitely into the past or the future. One example is a British bond of the 1910.s that pays interest .in perpetuity.. <> primordial Definition: time interval that begins each time interval Description: This is the earliest possible time interval, one that is before or starts every other time interval. Note: This concept can be used in formulations such as .primordial through today. to define time intervals that began at some indefinite time in the past. Tools may choose to support a convenient syntax such as .until today.. Example: .Primordial through Tuesday. meaning .until Tuesday.. perpetuity Definition: time interval that ends each time interval Description: This is the latest possible time interval, one that is after or ends every other time interval. Note: This concept can be used in formulations such as .2012 through perpetuity. to define time intervals that extend indefinitely into the future. Tools may choose to support a convenient syntax such as .after 2012.. Example: .Contract signing through perpetuity. meaning .after the contract signing.. eternity Synonym: forever Definition: primordial through perpetuity Description: The time interval that extends across the entire Time Axis. Note: Eternity is not the same thing as the Time Axis, even though it .covers. the Time Axis. In clause 8.6, replace figure 8.19 with this version, which deletes the instance .forever.. <> In clause 8.6, delete the glossary entry for .forever.. Disposition: Resolved To: date-time-ftf@omg.org Subject: Date Time Issue 17540 - Need to support infinite and indefinite time constructs X-KeepSent: 2E35D1A9:AA442D0C-85257A7C:004C327E; type=4; name=$KeepSent X-Mailer: Lotus Notes Release 8.5.3 September 15, 2011 From: Mark H Linehan Date: Mon, 17 Sep 2012 09:53:30 -0400 X-MIMETrack: Serialize by Router on D01MC604/01/M/IBM(Release 8.5.3FP2IF1|July 25, 2012) at 09/17/2012 09:53:31 x-cbid: 12091713-3534-0000-0000-00000CCF9EA8 This proposed resolution is an update based on last Thursday's discussion. I have added it to the agenda for today's conference call. ----------------------------- Mark H. Linehan STSM, IBM Research Date-Time Issue 17540 - Need to Support Infinite and Indefinite Time Constructs1.doc Disposition: Resolved OMG Issue No: 17540 Title: Need to support infinite and indefinite time constructs Source: Mike Bennett, EDM Council, mbennett@edmcouncil.org Summary: The Date Time Vocabulary needs to be able to support statements about periods of time which extend indefinitely into the future, and also describe periods of time which will have begun at indeterminate times in the past. As an example of the former, it is possible and meaningful for a contract to make statements about commitments or rights which extend in perpetuity, such as Perpetual Bonds which are bonds that pay interest forever. In general, it is necessary to be able to make meaningful statements which embody the concept of Forever. Similarly, it is necessary to be able to make meaningful statements which have been going on in perpetuity up to the present time. The DTV specification does currently allow for making statements about infinite time going forward, but not about time periods which have started at some indefinite time in the past. Meanwhile there are three other issues in play which touch on this same area. These may have an impact on the current ability to make statements about infinite time into the future as well, depending on their final resolution. The issues which have a bearing or potential bearing on this matter are: - Issue 16992: Corollaries to Axiom D.4 in 8.2.3 are misstated; - Issue 16993: no syntax for indefinite time periods (date-time-ftf); - Issue 16997: forever is misdefined In summary, the requirement that needs to be met with the resolution of this and the above-referenced issues is: 1. Extend forever with two new concepts: a. Indeterminate time in the past b. Indefinite time in the future 2. Ensure that the concept forever can be adequately defined (per 16997) including with reference to the time axis; 3. That there is syntax for the specification of indeterminate time periods that began at some point in the past and last up until the present (per 16993) 4. That the restatement of the axiom and corollaries referenced in 16992 take account of the two concepts above (indeterminate time in the past and indefinite time in the future) Resolution: The FTF decided that there is no value in distinguishing .indefinite. from .infinite.. It chose to add new concepts that provide the basis for time intervals that extend indefinitely into the past or the future. New individual concepts .primordiality. and .perpetuity. are defined respectively as the earliest time interval and the latest time interval. .Eternity. (synonym .forever.) is defined as .primordiality through perpetuity.. This permits formulations such as .primordiality through today. and .2012 through perpetuity.. A tool might support these formulations with a syntax such as .. until today .. and .. from 2012 on ... Issue 16993 adds verb concepts such as .time interval until situation model.. .primordiality. and .perpetuity. can substitute for the .time interval. role to enable formulations such as .primordiality until the Industrial Revolution.. Revised Text: (These update instructions apply to the beta-2 specification). Add a new clause 8.1.8 at the end of clause 8.1. The new clause adds a figure, so the following figures of clause 8 must be renumbered. 8.1.8 Indefinite Time Intervals Indefinite time intervals provide the basis for describing time intervals that extend indefinitely into the past or the future. One example is a British bond of the 1910.s that pays interest .in perpetuity.. <> primordiality Definition: time interval that is before each time interval Description: This is the earliest possible time interval, one that is before every time interval. Note: This concept can be used in formulations such as .primordiality through today. to define time intervals that began at some indefinite time in the past. Tools may choose to support a convenient syntax such as .until today.. Note: 'Primordiality' is an individual concept because there can be only one such time interval. Example: .Primordiality through Tuesday. meaning .until Tuesday.. perpetuity Definition: time interval that is after each time interval Description: This is the latest possible time interval, one that is after every time interval. Note: This concept can be used in formulations such as .2012 through perpetuity. to define time intervals that extend indefinitely into the future. Tools may choose to support a convenient syntax such as .after 2012.. Note: 'Perpetuity' is an individual concept because there can be only one such time interval. Example: .Contract signing through perpetuity. meaning .after the contract signing.. eternity Synonym: forever Definition: primordiality through perpetuity Description: The time interval that extends across the entire Time Axis. Note: 'Eternity' is an individual concept because there can be only one such time interval. Note: Eternity is not the same thing as the Time Axis, even though it .covers. the Time Axis. In clause 8.6, replace figure 8.19 with this version, which deletes the instance .forever.. <> In clause 8.6, delete the glossary entry for .forever.. Disposition: Resolved Date: Tue, 18 Sep 2012 12:52:44 -0400 From: Ed Barkmeyer Reply-To: Organization: NIST User-Agent: Thunderbird 2.0.0.23 (Windows/20090812) To: Mark H Linehan , OMG DateTimeVoc FTF Subject: DTV Issue 17540: problem with the draft, question about approach X-NISTMEL-MailScanner-Information: Please contact postmaster@mel.nist.gov for more information X-NISTMEL-MailScanner-ID: q8IGqnGS001378 X-NISTMEL-MailScanner: Found to be clean X-NISTMEL-MailScanner-SpamCheck: X-NISTMEL-MailScanner-From: edbark@nist.gov X-NISTMEL-MailScanner-Watermark: 1348591972.16036@eNl9/wqN5d7x2FYh+yZ5SQ X-Spam-Status: No It occurred to me last night literally in the middle of the night that the idea of primodiality and perpetuity as proposed in 17540 is logically inconsistent with the anti-atomic axiom in some other resolved issue. The axiom says: every time interval has at least one proper part that is a time interval. Now, since perpetuity is said to be a time interval, it follows that it must have a proper part, but it cannot be after any proper part of itself. And therefore there is a time interval that is not perpetuity and is not before perpetuity. One possible fix is to change the definitions to eliminate all time intervals that are part of perpetuity and primordiality. But are these artifice time intervals the best way of describing the finite Time Axis? I think these things are "time instants" (of duration 0) in disguise. That is, they really have duration 0 and no proper parts. If the Time Axis is finite, then there is at least one time interval that is before or starts with every time interval. But there is more than one. There are intervals of every possible duration that all start with the Big Bang. And similar behavior occurs at the other end. So 'primordial interval' is a category of time interval, not an individual. And "all time before x" is the primordial interval that ends with x. Similarly, "in perpetuity after x" is the perpetual interval that starts with x. This is one possible alternative approach. On the other hand, what is needed, according to Issue 16993, is verb concepts to relate situations to indefinite time. Does that require this ontological commitment at all? An alternative approach is to specify just such verbs, e.g.: occurrence is in perpetuity after time interval Definition: the occurrence occurs throughout all time intervals that are after the time interval occurrence is at all times before time interval Definition: the occurrence occurs throughout all time intervals that are before the time interval occurrence is eternal Definition: the occurrence occurs throughout all time intervals I believe this approach was discussed. The problem with this approach is that such occurrences do not have an occurrence interval. On one side or the other there is no time interval that the occurrence does not occur within. That is, these things are "indefinite occurrences". And indefinite occurrences don't have a duration, either. But is that important? Is it a problem that 'all times before time interval1' is not a time interval? Is it important that 'all times before time interval' is a time interval? The underlying question is whether a concept like "the term of a contract" is always a time interval. So I think we are between a rock and a hard place here. One of the following commitments must be made: a) not all time intervals have a duration b) the Time Axis is finite c) not all occurrences (or contracts) have an occurrence interval or a duration I agree with the decision to go with (b). But upon reflection (above), I think the resolution to 17540 needs more than a patch. That is, I think primordial and perpetual are categories of time interval, and 'all time up to x', 'all time thru x', 'all time from x on' (aka 'from x in perpetuity') and 'all time after x', are verbs that characterize primordial and perpetual time intervals. And the binary fact types described above can be defined to relate occurrences to such time intervals. (BTW, I think the additions to the vocabulary in the 16993 proposal are good, but they don't address the issue directly; they only enable the issue to be addressed by using the artificial time intervals.) -Ed -- Edward J. Barkmeyer Email: edbark@nist.gov National Institute of Standards & Technology Manufacturing Systems Integration Division 100 Bureau Drive, Stop 8263 Tel: +1 301-975-3528 Gaithersburg, MD 20899-8263 Cel: +1 240-672-5800 "The opinions expressed above do not reflect consensus of NIST, and have not been reviewed by any Government authority." From: Mike Bennett To: "edbark@nist.gov" , Mark H Linehan , OMG DateTimeVoc FTF Subject: RE: DTV Issue 17540: problem with the draft, question about approach Thread-Topic: DTV Issue 17540: problem with the draft, question about approach Thread-Index: AQHNlb4eKe5Lli5Ex0OyJadacRU7YpeQVATA Date: Tue, 18 Sep 2012 17:07:12 +0000 Accept-Language: en-US X-MS-Has-Attach: X-MS-TNEF-Correlator: x-originating-ip: [86.178.116.52] The issue here is that DTV chooses to make a commitment to the philosophical notion that there is no such thing as a time point, and that therefore all things which are described as time points must decompose into time intervals. This is clearly not true of the time point at the beginning of time nor of the time point at the end of time. Also the ways in which people may choose to speak about time may not always be in accordance with the chosen philosophy or ontological commitment of DTV. Mike -- Mike Bennett Head of Semantics and Standards EDM Council Tel: +44 20 7917 9522 Cell: +44 7721 420 730 www.edmcouncil.org Semantics Repository: www.hypercube.co.uk/edmcouncil From: Ed Barkmeyer [mailto:edbark@nist.gov] Sent: Tuesday, September 18, 2012 12:53 PM To: Mark H Linehan; OMG DateTimeVoc FTF Subject: DTV Issue 17540: problem with the draft, question about approach It occurred to me last night literally in the middle of the night that the idea of primodiality and perpetuity as proposed in 17540 is logically inconsistent with the anti-atomic axiom in some other resolved issue. The axiom says: every time interval has at least one proper part that is a time interval. Now, since perpetuity is said to be a time interval, it follows that it must have a proper part, but it cannot be after any proper part of itself. And therefore there is a time interval that is not perpetuity and is not before perpetuity. One possible fix is to change the definitions to eliminate all time intervals that are part of perpetuity and primordiality. But are these artifice time intervals the best way of describing the finite Time Axis? I think these things are "time instants" (of duration 0) in disguise. That is, they really have duration 0 and no proper parts. If the Time Axis is finite, then there is at least one time interval that is before or starts with every time interval. But there is more than one. There are intervals of every possible duration that all start with the Big Bang. And similar behavior occurs at the other end. So 'primordial interval' is a category of time interval, not an individual. And "all time before x" is the primordial interval that ends with x. Similarly, "in perpetuity after x" is the perpetual interval that starts with x. This is one possible alternative approach. On the other hand, what is needed, according to Issue 16993, is verb concepts to relate situations to indefinite time. Does that require this ontological commitment at all? An alternative approach is to specify just such verbs, e.g.: occurrence is in perpetuity after time interval Definition: the occurrence occurs throughout all time intervals that are after the time interval occurrence is at all times before time interval Definition: the occurrence occurs throughout all time intervals that are before the time interval occurrence is eternal Definition: the occurrence occurs throughout all time intervals I believe this approach was discussed. The problem with this approach is that such occurrences do not have an occurrence interval. On one side or the other there is no time interval that the occurrence does not occur within. That is, these things are "indefinite occurrences". And indefinite occurrences don't have a duration, either. But is that important? Is it a problem that 'all times before time interval1' is not a time interval? Is it important that 'all times before time interval' is a time interval? The underlying question is whether a concept like "the term of a contract" is always a time interval. So I think we are between a rock and a hard place here. One of the following commitments must be made: a) not all time intervals have a duration b) the Time Axis is finite c) not all occurrences (or contracts) have an occurrence interval or a duration I agree with the decision to go with (b). But upon reflection (above), I think the resolution to 17540 needs more than a patch. That is, I think primordial and perpetual are categories of time interval, and 'all time up to x', 'all time thru x', 'all time from x on' (aka 'from x in perpetuity') and 'all time after x', are verbs that characterize primordial and perpetual time intervals. And the binary fact types described above can be defined to relate occurrences to such time intervals. (BTW, I think the additions to the vocabulary in the 16993 proposal are good, but they don't address the issue directly; they only enable the issue to be addressed by using the artificial time intervals.) -Ed -- Edward J. Barkmeyer Email: edbark@nist.gov National Institute of Standards & Technology Manufacturing Systems Integration Division 100 Bureau Drive, Stop 8263 Tel: +1 301-975-3528 Gaithersburg, MD 20899-8263 Cel: +1 240-672-5800 "The opinions expressed above do not reflect consensus of NIST, and have not been reviewed by any Government authority." To: Cc: OMG DateTimeVoc FTF Subject: Re: DTV Issue 17540: problem with the draft, question about approach X-KeepSent: 95872A60:E0FDF1BC-85257A7D:0064A6E7; type=4; name=$KeepSent X-Mailer: Lotus Notes Release 8.5.3 September 15, 2011 From: Mark H Linehan Date: Tue, 18 Sep 2012 14:36:36 -0400 X-MIMETrack: Serialize by Router on D01MC604/01/M/IBM(Release 8.5.3FP2IF1|July 25, 2012) at 09/18/2012 14:36:37, Serialize complete at 09/18/2012 14:36:37 x-cbid: 12091818-5930-0000-0000-00000C3BA25D Ed, I think the problem is with our current definition of 'primordiality': "time interval that is before each time interval that is not primordiality". I think it should be "time interval that starts before each time interval that is not primordiality". This latter definition allows 'primordiality' to have proper parts. A similar solution using 'ends after' would work for 'perpetuity'. I do not believe we are assuming a finite Time Axis with these definitions. We don't place these time intervals at the earliest/latest time intervals but without saying that they exist. Nor do we need to make any of the three ontological commitments that you list. ----------------------------- Mark H. Linehan STSM, IBM Research From: Ed Barkmeyer To: Mark H Linehan/Watson/IBM@IBMUS, OMG DateTimeVoc FTF , Date: 09/18/2012 12:54 PM Subject: DTV Issue 17540: problem with the draft, question about approach -------------------------------------------------------------------------------- It occurred to me last night literally in the middle of the night that the idea of primodiality and perpetuity as proposed in 17540 is logically inconsistent with the anti-atomic axiom in some other resolved issue. The axiom says: every time interval has at least one proper part that is a time interval. Now, since perpetuity is said to be a time interval, it follows that it must have a proper part, but it cannot be after any proper part of itself. And therefore there is a time interval that is not perpetuity and is not before perpetuity. One possible fix is to change the definitions to eliminate all time intervals that are part of perpetuity and primordiality. But are these artifice time intervals the best way of describing the finite Time Axis? I think these things are "time instants" (of duration 0) in disguise. That is, they really have duration 0 and no proper parts. If the Time Axis is finite, then there is at least one time interval that is before or starts with every time interval. But there is more than one. There are intervals of every possible duration that all start with the Big Bang. And similar behavior occurs at the other end. So 'primordial interval' is a category of time interval, not an individual. And "all time before x" is the primordial interval that ends with x. Similarly, "in perpetuity after x" is the perpetual interval that starts with x. This is one possible alternative approach. On the other hand, what is needed, according to Issue 16993, is verb concepts to relate situations to indefinite time. Does that require this ontological commitment at all? An alternative approach is to specify just such verbs, e.g.: occurrence is in perpetuity after time interval Definition: the occurrence occurs throughout all time intervals that are after the time interval occurrence is at all times before time interval Definition: the occurrence occurs throughout all time intervals that are before the time interval occurrence is eternal Definition: the occurrence occurs throughout all time intervals I believe this approach was discussed. The problem with this approach is that such occurrences do not have an occurrence interval. On one side or the other there is no time interval that the occurrence does not occur within. That is, these things are "indefinite occurrences". And indefinite occurrences don't have a duration, either. But is that important? Is it a problem that 'all times before time interval1' is not a time interval? Is it important that 'all times before time interval' is a time interval? The underlying question is whether a concept like "the term of a contract" is always a time interval. So I think we are between a rock and a hard place here. One of the following commitments must be made: a) not all time intervals have a duration b) the Time Axis is finite c) not all occurrences (or contracts) have an occurrence interval or a duration I agree with the decision to go with (b). But upon reflection (above), I think the resolution to 17540 needs more than a patch. That is, I think primordial and perpetual are categories of time interval, and 'all time up to x', 'all time thru x', 'all time from x on' (aka 'from x in perpetuity') and 'all time after x', are verbs that characterize primordial and perpetual time intervals. And the binary fact types described above can be defined to relate occurrences to such time intervals. (BTW, I think the additions to the vocabulary in the 16993 proposal are good, but they don't address the issue directly; they only enable the issue to be addressed by using the artificial time intervals.) -Ed -- Edward J. Barkmeyer Email: edbark@nist.gov National Institute of Standards & Technology Manufacturing Systems Integration Division 100 Bureau Drive, Stop 8263 Tel: +1 301-975-3528 Gaithersburg, MD 20899-8263 Cel: +1 240-672-5800 "The opinions expressed above do not reflect consensus of NIST, and have not been reviewed by any Government authority." To: Mike Bennett Cc: OMG DateTimeVoc FTF , "edbark@nist.gov" Subject: RE: DTV Issue 17540: problem with the draft, question about approach X-KeepSent: FED70C53:3746145D-85257A7D:00664064; type=4; name=$KeepSent X-Mailer: Lotus Notes Release 8.5.3 September 15, 2011 From: Mark H Linehan Date: Tue, 18 Sep 2012 14:41:01 -0400 X-MIMETrack: Serialize by Router on D01MC604/01/M/IBM(Release 8.5.3FP2IF1|July 25, 2012) at 09/18/2012 14:41:02, Serialize complete at 09/18/2012 14:41:02 x-cbid: 12091818-7182-0000-0000-000002A6FA35 Mike, you are right that DTV "chooses to make a commitment to the philosophical notion that there is no such thing as a time point, and that therefore all things which are described as time points must decompose into time intervals." But the statement "This is clearly not true of the time point at the beginning of time nor of the time point at the end of time" seems to be a tautology. If you speak of a time point at the beginning of time, then you are necessarily thinking in terms of time points. I think we can show that DTV does not need to commit to a time point at the beginning of time. Of course people may have many different mental models of time. The DTV argument is that the idea of a point in time is not useful in business. ----------------------------- Mark H. Linehan STSM, IBM Research From: Mike Bennett To: "edbark@nist.gov" , Mark H Linehan/Watson/IBM@IBMUS, OMG DateTimeVoc FTF , Date: 09/18/2012 01:09 PM Subject: RE: DTV Issue 17540: problem with the draft, question about approach -------------------------------------------------------------------------------- The issue here is that DTV chooses to make a commitment to the philosophical notion that there is no such thing as a time point, and that therefore all things which are described as time points must decompose into time intervals. This is clearly not true of the time point at the beginning of time nor of the time point at the end of time. Also the ways in which people may choose to speak about time may not always be in accordance with the chosen philosophy or ontological commitment of DTV. Mike -- Mike Bennett Head of Semantics and Standards EDM Council Tel: +44 20 7917 9522 Cell: +44 7721 420 730 www.edmcouncil.org Semantics Repository: www.hypercube.co.uk/edmcouncil From: Ed Barkmeyer [mailto:edbark@nist.gov] Sent: Tuesday, September 18, 2012 12:53 PM To: Mark H Linehan; OMG DateTimeVoc FTF Subject: DTV Issue 17540: problem with the draft, question about approach It occurred to me last night literally in the middle of the night that the idea of primodiality and perpetuity as proposed in 17540 is logically inconsistent with the anti-atomic axiom in some other resolved issue. The axiom says: every time interval has at least one proper part that is a time interval. Now, since perpetuity is said to be a time interval, it follows that it must have a proper part, but it cannot be after any proper part of itself. And therefore there is a time interval that is not perpetuity and is not before perpetuity. One possible fix is to change the definitions to eliminate all time intervals that are part of perpetuity and primordiality. But are these artifice time intervals the best way of describing the finite Time Axis? I think these things are "time instants" (of duration 0) in disguise. That is, they really have duration 0 and no proper parts. If the Time Axis is finite, then there is at least one time interval that is before or starts with every time interval. But there is more than one. There are intervals of every possible duration that all start with the Big Bang. And similar behavior occurs at the other end. So 'primordial interval' is a category of time interval, not an individual. And "all time before x" is the primordial interval that ends with x. Similarly, "in perpetuity after x" is the perpetual interval that starts with x. This is one possible alternative approach. On the other hand, what is needed, according to Issue 16993, is verb concepts to relate situations to indefinite time. Does that require this ontological commitment at all? An alternative approach is to specify just such verbs, e.g.: occurrence is in perpetuity after time interval Definition: the occurrence occurs throughout all time intervals that are after the time interval occurrence is at all times before time interval Definition: the occurrence occurs throughout all time intervals that are before the time interval occurrence is eternal Definition: the occurrence occurs throughout all time intervals I believe this approach was discussed. The problem with this approach is that such occurrences do not have an occurrence interval. On one side or the other there is no time interval that the occurrence does not occur within. That is, these things are "indefinite occurrences". And indefinite occurrences don't have a duration, either. But is that important? Is it a problem that 'all times before time interval1' is not a time interval? Is it important that 'all times before time interval' is a time interval? The underlying question is whether a concept like "the term of a contract" is always a time interval. So I think we are between a rock and a hard place here. One of the following commitments must be made: a) not all time intervals have a duration b) the Time Axis is finite c) not all occurrences (or contracts) have an occurrence interval or a duration I agree with the decision to go with (b). But upon reflection (above), I think the resolution to 17540 needs more than a patch. That is, I think primordial and perpetual are categories of time interval, and 'all time up to x', 'all time thru x', 'all time from x on' (aka 'from x in perpetuity') and 'all time after x', are verbs that characterize primordial and perpetual time intervals. And the binary fact types described above can be defined to relate occurrences to such time intervals. (BTW, I think the additions to the vocabulary in the 16993 proposal are good, but they don't address the issue directly; they only enable the issue to be addressed by using the artificial time intervals.) -Ed -- Edward J. Barkmeyer Email: edbark@nist.gov National Institute of Standards & Technology Manufacturing Systems Integration Division 100 Bureau Drive, Stop 8263 Tel: +1 301-975-3528 Gaithersburg, MD 20899-8263 Cel: +1 240-672-5800 "The opinions expressed above do not reflect consensus of NIST, and have not been reviewed by any Government authority." From: Mike Bennett To: Mark H Linehan CC: OMG DateTimeVoc FTF , "edbark@nist.gov" Subject: RE: DTV Issue 17540: problem with the draft, question about approach Thread-Topic: DTV Issue 17540: problem with the draft, question about approach Thread-Index: AQHNlb4eKe5Lli5Ex0OyJadacRU7YpeQVATAgACQJYD//4xnIA== Date: Tue, 18 Sep 2012 18:49:02 +0000 Accept-Language: en-US X-MS-Has-Attach: X-MS-TNEF-Correlator: x-originating-ip: [86.178.116.52] Hi Mark, Yes, my point was that in talking of the time point at the beginning of time we are necessarily thinking in time points. But that these are not time points which decompose into time intervals. Mike -- Mike Bennett Head of Semantics and Standards EDM Council Tel: +44 20 7917 9522 Cell: +44 7721 420 730 www.edmcouncil.org Semantics Repository: www.hypercube.co.uk/edmcouncil From: Mark H Linehan [mailto:mlinehan@us.ibm.com] Sent: Tuesday, September 18, 2012 2:41 PM To: Mike Bennett Cc: OMG DateTimeVoc FTF; edbark@nist.gov Subject: RE: DTV Issue 17540: problem with the draft, question about approach Mike, you are right that DTV "chooses to make a commitment to the philosophical notion that there is no such thing as a time point, and that therefore all things which are described as time points must decompose into time intervals." But the statement "This is clearly not true of the time point at the beginning of time nor of the time point at the end of time" seems to be a tautology. If you speak of a time point at the beginning of time, then you are necessarily thinking in terms of time points. I think we can show that DTV does not need to commit to a time point at the beginning of time. Of course people may have many different mental models of time. The DTV argument is that the idea of a point in time is not useful in business. ----------------------------- Mark H. Linehan STSM, IBM Research From: Mike Bennett To: "edbark@nist.gov" , Mark H Linehan/Watson/IBM@IBMUS, OMG DateTimeVoc FTF , Date: 09/18/2012 01:09 PM Subject: RE: DTV Issue 17540: problem with the draft, question about approach -------------------------------------------------------------------------------- The issue here is that DTV chooses to make a commitment to the philosophical notion that there is no such thing as a time point, and that therefore all things which are described as time points must decompose into time intervals. This is clearly not true of the time point at the beginning of time nor of the time point at the end of time. Also the ways in which people may choose to speak about time may not always be in accordance with the chosen philosophy or ontological commitment of DTV. Mike -- Mike Bennett Head of Semantics and Standards EDM Council Tel: +44 20 7917 9522 Cell: +44 7721 420 730 www.edmcouncil.org Semantics Repository: www.hypercube.co.uk/edmcouncil From: Ed Barkmeyer [mailto:edbark@nist.gov] Sent: Tuesday, September 18, 2012 12:53 PM To: Mark H Linehan; OMG DateTimeVoc FTF Subject: DTV Issue 17540: problem with the draft, question about approach It occurred to me last night literally in the middle of the night that the idea of primodiality and perpetuity as proposed in 17540 is logically inconsistent with the anti-atomic axiom in some other resolved issue. The axiom says: every time interval has at least one proper part that is a time interval. Now, since perpetuity is said to be a time interval, it follows that it must have a proper part, but it cannot be after any proper part of itself. And therefore there is a time interval that is not perpetuity and is not before perpetuity. One possible fix is to change the definitions to eliminate all time intervals that are part of perpetuity and primordiality. But are these artifice time intervals the best way of describing the finite Time Axis? I think these things are "time instants" (of duration 0) in disguise. That is, they really have duration 0 and no proper parts. If the Time Axis is finite, then there is at least one time interval that is before or starts with every time interval. But there is more than one. There are intervals of every possible duration that all start with the Big Bang. And similar behavior occurs at the other end. So 'primordial interval' is a category of time interval, not an individual. And "all time before x" is the primordial interval that ends with x. Similarly, "in perpetuity after x" is the perpetual interval that starts with x. This is one possible alternative approach. On the other hand, what is needed, according to Issue 16993, is verb concepts to relate situations to indefinite time. Does that require this ontological commitment at all? An alternative approach is to specify just such verbs, e.g.: occurrence is in perpetuity after time interval Definition: the occurrence occurs throughout all time intervals that are after the time interval occurrence is at all times before time interval Definition: the occurrence occurs throughout all time intervals that are before the time interval occurrence is eternal Definition: the occurrence occurs throughout all time intervals I believe this approach was discussed. The problem with this approach is that such occurrences do not have an occurrence interval. On one side or the other there is no time interval that the occurrence does not occur within. That is, these things are "indefinite occurrences". And indefinite occurrences don't have a duration, either. But is that important? Is it a problem that 'all times before time interval1' is not a time interval? Is it important that 'all times before time interval' is a time interval? The underlying question is whether a concept like "the term of a contract" is always a time interval. So I think we are between a rock and a hard place here. One of the following commitments must be made: a) not all time intervals have a duration b) the Time Axis is finite c) not all occurrences (or contracts) have an occurrence interval or a duration I agree with the decision to go with (b). But upon reflection (above), I think the resolution to 17540 needs more than a patch. That is, I think primordial and perpetual are categories of time interval, and 'all time up to x', 'all time thru x', 'all time from x on' (aka 'from x in perpetuity') and 'all time after x', are verbs that characterize primordial and perpetual time intervals. And the binary fact types described above can be defined to relate occurrences to such time intervals. (BTW, I think the additions to the vocabulary in the 16993 proposal are good, but they don't address the issue directly; they only enable the issue to be addressed by using the artificial time intervals.) -Ed -- Edward J. Barkmeyer Email: edbark@nist.gov National Institute of Standards & Technology Manufacturing Systems Integration Division 100 Bureau Drive, Stop 8263 Tel: +1 301-975-3528 Gaithersburg, MD 20899-8263 Cel: +1 240-672-5800 "The opinions expressed above do not reflect consensus of NIST, and have not been reviewed by any Government authority." Date: Tue, 18 Sep 2012 18:57:26 -0400 From: Edward Barkmeyer Reply-To: Organization: NIST User-Agent: Thunderbird 2.0.0.23 (Windows/20090812) To: Mark H Linehan CC: OMG DateTimeVoc FTF Subject: Re: DTV Issue 17540: problem with the draft, question about approach Mark H Linehan wrote: Ed, I think the problem is with our current definition of 'primordiality': "time interval that is before each time interval that is not primordiality". I think it should be "time interval that starts before each time interval that is not primordiality". This latter definition allows 'primordiality' to have proper parts. OK. I like that fix. But with that definition, it does not seem that 'primordiality' is an individual concept. It seems to be a category of time interval, and thus should read "each time interval that is not A primordiality", but that definition might then be considered circular. If t is a primordiality, there is a proper part of t that starts t, and that proper part is then also a primordiality. Note that I suggested the definition: time interval that starts with or before each time interval. But that was ambiguous. What was meant is 'time interval such that for each time interval1, either time interval starts with time interval1 or time interval starts before time interval1. Every time interval that it starts with is also a primordial time interval, by definition. A similar solution using 'ends after' would work for 'perpetuity'. I do not believe we are assuming a finite Time Axis with these definitions. You are right. What I meant by 'finite' is that the Time Axis is itself a time interval that has a fixed duration. And that does not follow. What I was trying to say is that the Time Axis is bounded above and below as a space of time intervals under the ordering 'before'. We don't place these time intervals at the earliest/latest time intervals but without saying that they exist. I don't quite understand. The definitions do make them earliest and latest. You do not intend to assert that the category 'perpetuity' has an instance? If not, then a reference to 'January 1,2013 through perpetuity' doesn't necessarily have a meaning. Nor do we need to make any of the three ontological commitments that you list. I agree that if you don't insist that perpetuity has a referent, then you make no ontological commitment. Presumably, however, anyone who wants to use 'perpetuity' in a contract must make that commitment. Is that the intent? If we are being so careful as that, I think it would be appropriate to add a Note to that effect. Mike Bennett wrote: Yes, my point was that in talking of the time point at the beginning of time we are necessarily thinking in time points. But that these are not time points which decompose into time intervals. I see that you were thinking in terms of time "instants" (time intervals of duration 0). My point was that the concept of perpetuity doesn't require a 'time instant' at the "end of time", or even a "dies irae" (which has duration 1 day?). It can be thought of as "any time interval that ends at the end of time", e.g., "from the signing of the contract forward". Or it can be thought of as "all time intervals after some specified point in time", like the signing of a contract. Apart from Gabriel's trumpet blast, we probably do not need to speak of things that occur "AT the end of time"; we only need to talk about things that occur over intervals that end at the end of time, or occur over all time intervals after some point. -Ed -- Edward J. Barkmeyer Email: edbark@nist.gov National Institute of Standards & Technology Manufacturing Systems Integration Division 100 Bureau Drive, Stop 8263 Tel: +1 301-975-3528 Gaithersburg, MD 20899-8263 Cel: +1 240-672-5800 "The opinions expressed above do not reflect consensus of NIST, and have not been reviewed by any Government authority." ----------------------------- Mark H. Linehan STSM, IBM Research From: Ed Barkmeyer To: Mark H Linehan/Watson/IBM@IBMUS, OMG DateTimeVoc FTF , Date: 09/18/2012 12:54 PM Subject: DTV Issue 17540: problem with the draft, question about approach -------------------------------------------------------------------------------- It occurred to me last night literally in the middle of the night that the idea of primodiality and perpetuity as proposed in 17540 is logically inconsistent with the anti-atomic axiom in some other resolved issue. The axiom says: every time interval has at least one proper part that is a time interval. Now, since perpetuity is said to be a time interval, it follows that it must have a proper part, but it cannot be after any proper part of itself. And therefore there is a time interval that is not perpetuity and is not before perpetuity. One possible fix is to change the definitions to eliminate all time intervals that are part of perpetuity and primordiality. But are these artifice time intervals the best way of describing the finite Time Axis? I think these things are "time instants" (of duration 0) in disguise. That is, they really have duration 0 and no proper parts. If the Time Axis is finite, then there is at least one time interval that is before or starts with every time interval. But there is more than one. There are intervals of every possible duration that all start with the Big Bang. And similar behavior occurs at the other end. So 'primordial interval' is a category of time interval, not an individual. And "all time before x" is the primordial interval that ends with x. Similarly, "in perpetuity after x" is the perpetual interval that starts with x. This is one possible alternative approach. On the other hand, what is needed, according to Issue 16993, is verb concepts to relate situations to indefinite time. Does that require this ontological commitment at all? An alternative approach is to specify just such verbs, e.g.: occurrence is in perpetuity after time interval Definition: the occurrence occurs throughout all time intervals that are after the time interval occurrence is at all times before time interval Definition: the occurrence occurs throughout all time intervals that are before the time interval occurrence is eternal Definition: the occurrence occurs throughout all time intervals I believe this approach was discussed. The problem with this approach is that such occurrences do not have an occurrence interval. On one side or the other there is no time interval that the occurrence does not occur within. That is, these things are "indefinite occurrences". And indefinite occurrences don't have a duration, either. But is that important? Is it a problem that 'all times before time interval1' is not a time interval? Is it important that 'all times before time interval' is a time interval? The underlying question is whether a concept like "the term of a contract" is always a time interval. So I think we are between a rock and a hard place here. One of the following commitments must be made: a) not all time intervals have a duration b) the Time Axis is finite c) not all occurrences (or contracts) have an occurrence interval or a duration I agree with the decision to go with (b). But upon reflection (above), I think the resolution to 17540 needs more than a patch. That is, I think primordial and perpetual are categories of time interval, and 'all time up to x', 'all time thru x', 'all time from x on' (aka 'from x in perpetuity') and 'all time after x', are verbs that characterize primordial and perpetual time intervals. And the binary fact types described above can be defined to relate occurrences to such time intervals. (BTW, I think the additions to the vocabulary in the 16993 proposal are good, but they don't address the issue directly; they only enable the issue to be addressed by using the artificial time intervals.) -Ed -- Edward J. Barkmeyer Email: edbark@nist.gov National Institute of Standards & Technology Manufacturing Systems Integration Division 100 Bureau Drive, Stop 8263 Tel: +1 301-975-3528 Gaithersburg, MD 20899-8263 Cel: +1 240-672-5800 "The opinions expressed above do not reflect consensus of NIST, and have not been reviewed by any Government authority." To: Cc: OMG DateTimeVoc FTF Subject: Re: DTV Issue 17540: problem with the draft, question about approach X-KeepSent: 7A5AB9E7:7FAECC96-85257A7E:00488392; type=4; name=$KeepSent X-Mailer: Lotus Notes Release 8.5.3 September 15, 2011 From: Mark H Linehan Date: Wed, 19 Sep 2012 09:22:32 -0400 X-MIMETrack: Serialize by Router on D01MC604/01/M/IBM(Release 8.5.3FP2IF1|July 25, 2012) at 09/19/2012 09:22:38, Serialize complete at 09/19/2012 09:22:38 x-cbid: 12091913-5806-0000-0000-000019C1FF15 Ed, I don't understand why my proposed definition of 'primordiality' defines a category rather than an individual concept. I think there is a maximum of one time interval that starts before every other time interval. It seems to me that 'primordiality' starts before any time interval t that is a proper part of 'primordiality'. As defined in clause 8.1.3, 'starts before' subsumes 'starts'. So there is no reason to use "... starts with ... or starts before ...." I think a reference to 'January 1,2013 through perpetuity' need not commit one to the idea that there is a 'perpetuity' time interval. It just commits the time interval to start at the given date and keep going. If there is a perpetuity then the time interval ends with it. And if there is no perpetuity, then the time interval just continues. Maybe it will reach 'perpetuity' some day -- but that is not for any of us to know. I have no objection to adding another note about that. ----------------------------- Mark H. Linehan STSM, IBM Research From: Edward Barkmeyer To: Mark H Linehan/Watson/IBM@IBMUS, Cc: OMG DateTimeVoc FTF Date: 09/18/2012 06:57 PM Subject: Re: DTV Issue 17540: problem with the draft, question about approach -------------------------------------------------------------------------------- Mark H Linehan wrote: Ed, I think the problem is with our current definition of 'primordiality': "time interval that is before each time interval that is not primordiality". I think it should be "time interval that starts before each time interval that is not primordiality". This latter definition allows 'primordiality' to have proper parts. OK. I like that fix. But with that definition, it does not seem that 'primordiality' is an individual concept. It seems to be a category of time interval, and thus should read "each time interval that is not A primordiality", but that definition might then be considered circular. If t is a primordiality, there is a proper part of t that starts t, and that proper part is then also a primordiality. Note that I suggested the definition: time interval that starts with or before each time interval. But that was ambiguous. What was meant is 'time interval such that for each time interval1, either time interval starts with time interval1 or time interval starts before time interval1. Every time interval that it starts with is also a primordial time interval, by definition. A similar solution using 'ends after' would work for 'perpetuity'. I do not believe we are assuming a finite Time Axis with these definitions. You are right. What I meant by 'finite' is that the Time Axis is itself a time interval that has a fixed duration. And that does not follow. What I was trying to say is that the Time Axis is bounded above and below as a space of time intervals under the ordering 'before'. We don't place these time intervals at the earliest/latest time intervals but without saying that they exist. I don't quite understand. The definitions do make them earliest and latest. You do not intend to assert that the category 'perpetuity' has an instance? If not, then a reference to 'January 1,2013 through perpetuity' doesn't necessarily have a meaning. Nor do we need to make any of the three ontological commitments that you list. I agree that if you don't insist that perpetuity has a referent, then you make no ontological commitment. Presumably, however, anyone who wants to use 'perpetuity' in a contract must make that commitment. Is that the intent? If we are being so careful as that, I think it would be appropriate to add a Note to that effect. Mike Bennett wrote: Yes, my point was that in talking of the time point at the beginning of time we are necessarily thinking in time points. But that these are not time points which decompose into time intervals. I see that you were thinking in terms of time "instants" (time intervals of duration 0). My point was that the concept of perpetuity doesn't require a 'time instant' at the "end of time", or even a "dies irae" (which has duration 1 day?). It can be thought of as "any time interval that ends at the end of time", e.g., "from the signing of the contract forward". Or it can be thought of as "all time intervals after some specified point in time", like the signing of a contract. Apart from Gabriel's trumpet blast, we probably do not need to speak of things that occur "AT the end of time"; we only need to talk about things that occur over intervals that end at the end of time, or occur over all time intervals after some point. -Ed -- Edward J. Barkmeyer Email: edbark@nist.gov National Institute of Standards & Technology Manufacturing Systems Integration Division 100 Bureau Drive, Stop 8263 Tel: +1 301-975-3528 Gaithersburg, MD 20899-8263 Cel: +1 240-672-5800 "The opinions expressed above do not reflect consensus of NIST, and have not been reviewed by any Government authority." ----------------------------- Mark H. Linehan STSM, IBM Research From: Ed Barkmeyer To: Mark H Linehan/Watson/IBM@IBMUS, OMG DateTimeVoc FTF , Date: 09/18/2012 12:54 PM Subject: DTV Issue 17540: problem with the draft, question about approach -------------------------------------------------------------------------------- It occurred to me last night literally in the middle of the night that the idea of primodiality and perpetuity as proposed in 17540 is logically inconsistent with the anti-atomic axiom in some other resolved issue. The axiom says: every time interval has at least one proper part that is a time interval. Now, since perpetuity is said to be a time interval, it follows that it must have a proper part, but it cannot be after any proper part of itself. And therefore there is a time interval that is not perpetuity and is not before perpetuity. One possible fix is to change the definitions to eliminate all time intervals that are part of perpetuity and primordiality. But are these artifice time intervals the best way of describing the finite Time Axis? I think these things are "time instants" (of duration 0) in disguise. That is, they really have duration 0 and no proper parts. If the Time Axis is finite, then there is at least one time interval that is before or starts with every time interval. But there is more than one. There are intervals of every possible duration that all start with the Big Bang. And similar behavior occurs at the other end. So 'primordial interval' is a category of time interval, not an individual. And "all time before x" is the primordial interval that ends with x. Similarly, "in perpetuity after x" is the perpetual interval that starts with x. This is one possible alternative approach. On the other hand, what is needed, according to Issue 16993, is verb concepts to relate situations to indefinite time. Does that require this ontological commitment at all? An alternative approach is to specify just such verbs, e.g.: occurrence is in perpetuity after time interval Definition: the occurrence occurs throughout all time intervals that are after the time interval occurrence is at all times before time interval Definition: the occurrence occurs throughout all time intervals that are before the time interval occurrence is eternal Definition: the occurrence occurs throughout all time intervals I believe this approach was discussed. The problem with this approach is that such occurrences do not have an occurrence interval. On one side or the other there is no time interval that the occurrence does not occur within. That is, these things are "indefinite occurrences". And indefinite occurrences don't have a duration, either. But is that important? Is it a problem that 'all times before time interval1' is not a time interval? Is it important that 'all times before time interval' is a time interval? The underlying question is whether a concept like "the term of a contract" is always a time interval. So I think we are between a rock and a hard place here. One of the following commitments must be made: a) not all time intervals have a duration b) the Time Axis is finite c) not all occurrences (or contracts) have an occurrence interval or a duration I agree with the decision to go with (b). But upon reflection (above), I think the resolution to 17540 needs more than a patch. That is, I think primordial and perpetual are categories of time interval, and 'all time up to x', 'all time thru x', 'all time from x on' (aka 'from x in perpetuity') and 'all time after x', are verbs that characterize primordial and perpetual time intervals. And the binary fact types described above can be defined to relate occurrences to such time intervals. (BTW, I think the additions to the vocabulary in the 16993 proposal are good, but they don't address the issue directly; they only enable the issue to be addressed by using the artificial time intervals.) -Ed -- Edward J. Barkmeyer Email: edbark@nist.gov National Institute of Standards & Technology Manufacturing Systems Integration Division 100 Bureau Drive, Stop 8263 Tel: +1 301-975-3528 Gaithersburg, MD 20899-8263 Cel: +1 240-672-5800 "The opinions expressed above do not reflect consensus of NIST, and have not been reviewed by any Government authority." Date: Wed, 19 Sep 2012 11:51:49 -0400 From: Edward Barkmeyer Reply-To: Organization: NIST User-Agent: Thunderbird 2.0.0.23 (Windows/20090812) To: Mark H Linehan CC: OMG DateTimeVoc FTF Subject: Re: DTV Issue 17540: problem with the draft, question about approach Mark H Linehan wrote: Ed, I don't understand why my proposed definition of 'primordiality' defines a category rather than an individual concept. I think there is a maximum of one time interval that starts before every other time interval. It seems to me that 'primordiality' starts before any time interval t that is a proper part of 'primordiality'. The problem is exactly what Mike pointed out. The primordiality time interval has to have duration 0. Let it have duration one microsecond, and there is a time inteval that starts it and has duration 1 nanosecond. And primordiality does not start before the 1 nanosecond time interval, and that one does not start before the one that is 1 picosecond in duration, etc. There are no atoms, so the primordiality time interval has a proper part, and if the proper part does not start the primordiality, then there is a time interval that starts the primodiality and meets the proper part. As Mike said, this is all a consequence of the axioms of the interval model -- you can't really model an 'instant'. As defined in clause 8.1.3, 'starts before' subsumes 'starts'. So there is no reason to use "... starts with ... or starts before ...." As defined in 8.1.4, 'starts before' does not subsume 'starts'. Even if it did, it doesn't change the fact that primordiality is a general concept. All of the proper parts that start a primordial time interval are primordial time intervals by the given definition. I think a reference to 'January 1,2013 through perpetuity' need not commit one to the idea that there is a 'perpetuity' time interval. It just commits the time interval to start at the given date and keep going. No. If there is no 'perpetuity' time interval, then the expression 'January 1,2013 through perpetuity' is meaningless, or at best the proposition that includes it is false. If you say "the King of the USA was married on 5 September 2002", what does it mean? It is either meaningless or false, because "the King of the USA" has no referent. If there is a perpetuity then the time interval ends with it. And if there is no perpetuity, then the time interval just continues. If there is no perpetuity, then 'January 1,2013 through perpetuity' does not refer to a time interval. Maybe it will reach 'perpetuity' some day -- but that is not for any of us to know. I have no objection to adding another note about that. The note will say that we are trying to make a silk purse out of a sow's ear and no pig was harmed in constructing it. The proposed model requires the ontological commitment to 'the end of time'. -Ed ----------------------------- Mark H. Linehan STSM, IBM Research From: Edward Barkmeyer To: Mark H Linehan/Watson/IBM@IBMUS, Cc: OMG DateTimeVoc FTF Date: 09/18/2012 06:57 PM Subject: Re: DTV Issue 17540: problem with the draft, question about approach -------------------------------------------------------------------------------- Mark H Linehan wrote: Ed, I think the problem is with our current definition of 'primordiality': "time interval that is before each time interval that is not primordiality". I think it should be "time interval that starts before each time interval that is not primordiality". This latter definition allows 'primordiality' to have proper parts. OK. I like that fix. But with that definition, it does not seem that 'primordiality' is an individual concept. It seems to be a category of time interval, and thus should read "each time interval that is not A primordiality", but that definition might then be considered circular. If t is a primordiality, there is a proper part of t that starts t, and that proper part is then also a primordiality. Note that I suggested the definition: time interval that starts with or before each time interval. But that was ambiguous. What was meant is 'time interval such that for each time interval1, either time interval starts with time interval1 or time interval starts before time interval1. Every time interval that it starts with is also a primordial time interval, by definition. A similar solution using 'ends after' would work for 'perpetuity'. I do not believe we are assuming a finite Time Axis with these definitions. You are right. What I meant by 'finite' is that the Time Axis is itself a time interval that has a fixed duration. And that does not follow. What I was trying to say is that the Time Axis is bounded above and below as a space of time intervals under the ordering 'before'. We don't place these time intervals at the earliest/latest time intervals but without saying that they exist. I don't quite understand. The definitions do make them earliest and latest. You do not intend to assert that the category 'perpetuity' has an instance? If not, then a reference to 'January 1,2013 through perpetuity' doesn't necessarily have a meaning. Nor do we need to make any of the three ontological commitments that you list. I agree that if you don't insist that perpetuity has a referent, then you make no ontological commitment. Presumably, however, anyone who wants to use 'perpetuity' in a contract must make that commitment. Is that the intent? If we are being so careful as that, I think it would be appropriate to add a Note to that effect. Mike Bennett wrote: Yes, my point was that in talking of the time point at the beginning of time we are necessarily thinking in time points. But that these are not time points which decompose into time intervals. I see that you were thinking in terms of time "instants" (time intervals of duration 0). My point was that the concept of perpetuity doesn't require a 'time instant' at the "end of time", or even a "dies irae" (which has duration 1 day?). It can be thought of as "any time interval that ends at the end of time", e.g., "from the signing of the contract forward". Or it can be thought of as "all time intervals after some specified point in time", like the signing of a contract. Apart from Gabriel's trumpet blast, we probably do not need to speak of things that occur "AT the end of time"; we only need to talk about things that occur over intervals that end at the end of time, or occur over all time intervals after some point. -Ed -- Edward J. Barkmeyer Email: edbark@nist.gov National Institute of Standards & Technology Manufacturing Systems Integration Division 100 Bureau Drive, Stop 8263 Tel: +1 301-975-3528 Gaithersburg, MD 20899-8263 Cel: +1 240-672-5800 "The opinions expressed above do not reflect consensus of NIST, and have not been reviewed by any Government authority." ----------------------------- Mark H. Linehan STSM, IBM Research From: Ed Barkmeyer To: Mark H Linehan/Watson/IBM@IBMUS, OMG DateTimeVoc FTF , Date: 09/18/2012 12:54 PM Subject: DTV Issue 17540: problem with the draft, question about approach -------------------------------------------------------------------------------- It occurred to me last night literally in the middle of the night that the idea of primodiality and perpetuity as proposed in 17540 is logically inconsistent with the anti-atomic axiom in some other resolved issue. The axiom says: every time interval has at least one proper part that is a time interval. Now, since perpetuity is said to be a time interval, it follows that it must have a proper part, but it cannot be after any proper part of itself. And therefore there is a time interval that is not perpetuity and is not before perpetuity. One possible fix is to change the definitions to eliminate all time intervals that are part of perpetuity and primordiality. But are these artifice time intervals the best way of describing the finite Time Axis? I think these things are "time instants" (of duration 0) in disguise. That is, they really have duration 0 and no proper parts. If the Time Axis is finite, then there is at least one time interval that is before or starts with every time interval. But there is more than one. There are intervals of every possible duration that all start with the Big Bang. And similar behavior occurs at the other end. So 'primordial interval' is a category of time interval, not an individual. And "all time before x" is the primordial interval that ends with x. Similarly, "in perpetuity after x" is the perpetual interval that starts with x. This is one possible alternative approach. On the other hand, what is needed, according to Issue 16993, is verb concepts to relate situations to indefinite time. Does that require this ontological commitment at all? An alternative approach is to specify just such verbs, e.g.: occurrence is in perpetuity after time interval Definition: the occurrence occurs throughout all time intervals that are after the time interval occurrence is at all times before time interval Definition: the occurrence occurs throughout all time intervals that are before the time interval occurrence is eternal Definition: the occurrence occurs throughout all time intervals I believe this approach was discussed. The problem with this approach is that such occurrences do not have an occurrence interval. On one side or the other there is no time interval that the occurrence does not occur within. That is, these things are "indefinite occurrences". And indefinite occurrences don't have a duration, either. But is that important? Is it a problem that 'all times before time interval1' is not a time interval? Is it important that 'all times before time interval' is a time interval? The underlying question is whether a concept like "the term of a contract" is always a time interval. So I think we are between a rock and a hard place here. One of the following commitments must be made: a) not all time intervals have a duration b) the Time Axis is finite c) not all occurrences (or contracts) have an occurrence interval or a duration I agree with the decision to go with (b). But upon reflection (above), I think the resolution to 17540 needs more than a patch. That is, I think primordial and perpetual are categories of time interval, and 'all time up to x', 'all time thru x', 'all time from x on' (aka 'from x in perpetuity') and 'all time after x', are verbs that characterize primordial and perpetual time intervals. And the binary fact types described above can be defined to relate occurrences to such time intervals. (BTW, I think the additions to the vocabulary in the 16993 proposal are good, but they don't address the issue directly; they only enable the issue to be addressed by using the artificial time intervals.) -Ed -- Edward J. Barkmeyer Email: edbark@nist.gov National Institute of Standards & Technology Manufacturing Systems Integration Division 100 Bureau Drive, Stop 8263 Tel: +1 301-975-3528 Gaithersburg, MD 20899-8263 Cel: +1 240-672-5800 "The opinions expressed above do not reflect consensus of NIST, and have not been reviewed by any Government authority." -- Edward J. Barkmeyer Email: edbark@nist.gov National Institute of Standards & Technology Manufacturing Systems Integration Division 100 Bureau Drive, Stop 8263 Tel: +1 301-975-3528 Gaithersburg, MD 20899-8263 Cel: +1 240-672-5800 "The opinions expressed above do not reflect consensus of NIST, and have not been reviewed by any Government authority." To: Cc: OMG DateTimeVoc FTF Subject: Re: DTV Issue 17540: problem with the draft, question about approach X-KeepSent: 2081135A:BAC036FF-85257A7F:00412B39; type=4; name=$KeepSent X-Mailer: Lotus Notes Release 8.5.3 September 15, 2011 From: Mark H Linehan Date: Thu, 20 Sep 2012 07:58:21 -0400 X-MIMETrack: Serialize by Router on D01MC604/01/M/IBM(Release 8.5.3FP2IF1|July 25, 2012) at 09/20/2012 07:58:24, Serialize complete at 09/20/2012 07:58:24 x-cbid: 12092011-6078-0000-0000-00000FC86FB0 Ed, I'm ok with your proposed definition: 'time interval that for each time interval1, either the time interval starts the time interval1 or the time interval starts before the time interval1'. With this definition, I think 'primordiality' is an individual concept. Regarding 'January 1,2013 through perpetuity' -- it seems to me that we cannot know whether 'perpetuity' represents a time interval or not, at least not until we reach it. By specifying 'perpetuity' as an individual concept, we avoid committing to that choice. This correctly models the world that we experience. We name something so that we can talk about it, even though we don't know now, and may never know, whether the individual concept has an instance. ----------------------------- Mark H. Linehan STSM, IBM Research From: Edward Barkmeyer To: Mark H Linehan/Watson/IBM@IBMUS, Cc: OMG DateTimeVoc FTF Date: 09/19/2012 11:53 AM Subject: Re: DTV Issue 17540: problem with the draft, question about approach -------------------------------------------------------------------------------- Mark H Linehan wrote: Ed, I don't understand why my proposed definition of 'primordiality' defines a category rather than an individual concept. I think there is a maximum of one time interval that starts before every other time interval. It seems to me that 'primordiality' starts before any time interval t that is a proper part of 'primordiality'. The problem is exactly what Mike pointed out. The primordiality time interval has to have duration 0. Let it have duration one microsecond, and there is a time inteval that starts it and has duration 1 nanosecond. And primordiality does not start before the 1 nanosecond time interval, and that one does not start before the one that is 1 picosecond in duration, etc. There are no atoms, so the primordiality time interval has a proper part, and if the proper part does not start the primordiality, then there is a time interval that starts the primodiality and meets the proper part. As Mike said, this is all a consequence of the axioms of the interval model -- you can't really model an 'instant'. As defined in clause 8.1.3, 'starts before' subsumes 'starts'. So there is no reason to use "... starts with ... or starts before ...." As defined in 8.1.4, 'starts before' does not subsume 'starts'. Even if it did, it doesn't change the fact that primordiality is a general concept. All of the proper parts that start a primordial time interval are primordial time intervals by the given definition. I think a reference to 'January 1,2013 through perpetuity' need not commit one to the idea that there is a 'perpetuity' time interval. It just commits the time interval to start at the given date and keep going. No. If there is no 'perpetuity' time interval, then the expression 'January 1,2013 through perpetuity' is meaningless, or at best the proposition that includes it is false. If you say "the King of the USA was married on 5 September 2002", what does it mean? It is either meaningless or false, because "the King of the USA" has no referent. If there is a perpetuity then the time interval ends with it. And if there is no perpetuity, then the time interval just continues. If there is no perpetuity, then 'January 1,2013 through perpetuity' does not refer to a time interval. Maybe it will reach 'perpetuity' some day -- but that is not for any of us to know. I have no objection to adding another note about that. The note will say that we are trying to make a silk purse out of a sow's ear and no pig was harmed in constructing it. The proposed model requires the ontological commitment to 'the end of time'. -Ed ----------------------------- Mark H. Linehan STSM, IBM Research From: Edward Barkmeyer To: Mark H Linehan/Watson/IBM@IBMUS, Cc: OMG DateTimeVoc FTF Date: 09/18/2012 06:57 PM Subject: Re: DTV Issue 17540: problem with the draft, question about approach -------------------------------------------------------------------------------- Mark H Linehan wrote: Ed, I think the problem is with our current definition of 'primordiality': "time interval that is before each time interval that is not primordiality". I think it should be "time interval that starts before each time interval that is not primordiality". This latter definition allows 'primordiality' to have proper parts. OK. I like that fix. But with that definition, it does not seem that 'primordiality' is an individual concept. It seems to be a category of time interval, and thus should read "each time interval that is not A primordiality", but that definition might then be considered circular. If t is a primordiality, there is a proper part of t that starts t, and that proper part is then also a primordiality. Note that I suggested the definition: time interval that starts with or before each time interval. But that was ambiguous. What was meant is 'time interval such that for each time interval1, either time interval starts with time interval1 or time interval starts before time interval1. Every time interval that it starts with is also a primordial time interval, by definition. A similar solution using 'ends after' would work for 'perpetuity'. I do not believe we are assuming a finite Time Axis with these definitions. You are right. What I meant by 'finite' is that the Time Axis is itself a time interval that has a fixed duration. And that does not follow. What I was trying to say is that the Time Axis is bounded above and below as a space of time intervals under the ordering 'before'. We don't place these time intervals at the earliest/latest time intervals but without saying that they exist. I don't quite understand. The definitions do make them earliest and latest. You do not intend to assert that the category 'perpetuity' has an instance? If not, then a reference to 'January 1,2013 through perpetuity' doesn't necessarily have a meaning. Nor do we need to make any of the three ontological commitments that you list. I agree that if you don't insist that perpetuity has a referent, then you make no ontological commitment. Presumably, however, anyone who wants to use 'perpetuity' in a contract must make that commitment. Is that the intent? If we are being so careful as that, I think it would be appropriate to add a Note to that effect. Mike Bennett wrote: Yes, my point was that in talking of the time point at the beginning of time we are necessarily thinking in time points. But that these are not time points which decompose into time intervals. I see that you were thinking in terms of time "instants" (time intervals of duration 0). My point was that the concept of perpetuity doesn't require a 'time instant' at the "end of time", or even a "dies irae" (which has duration 1 day?). It can be thought of as "any time interval that ends at the end of time", e.g., "from the signing of the contract forward". Or it can be thought of as "all time intervals after some specified point in time", like the signing of a contract. Apart from Gabriel's trumpet blast, we probably do not need to speak of things that occur "AT the end of time"; we only need to talk about things that occur over intervals that end at the end of time, or occur over all time intervals after some point. -Ed -- Edward J. Barkmeyer Email: edbark@nist.gov National Institute of Standards & Technology Manufacturing Systems Integration Division 100 Bureau Drive, Stop 8263 Tel: +1 301-975-3528 Gaithersburg, MD 20899-8263 Cel: +1 240-672-5800 "The opinions expressed above do not reflect consensus of NIST, and have not been reviewed by any Government authority." ----------------------------- Mark H. Linehan STSM, IBM Research From: Ed Barkmeyer To: Mark H Linehan/Watson/IBM@IBMUS, OMG DateTimeVoc FTF , Date: 09/18/2012 12:54 PM Subject: DTV Issue 17540: problem with the draft, question about approach -------------------------------------------------------------------------------- It occurred to me last night literally in the middle of the night that the idea of primodiality and perpetuity as proposed in 17540 is logically inconsistent with the anti-atomic axiom in some other resolved issue. The axiom says: every time interval has at least one proper part that is a time interval. Now, since perpetuity is said to be a time interval, it follows that it must have a proper part, but it cannot be after any proper part of itself. And therefore there is a time interval that is not perpetuity and is not before perpetuity. One possible fix is to change the definitions to eliminate all time intervals that are part of perpetuity and primordiality. But are these artifice time intervals the best way of describing the finite Time Axis? I think these things are "time instants" (of duration 0) in disguise. That is, they really have duration 0 and no proper parts. If the Time Axis is finite, then there is at least one time interval that is before or starts with every time interval. But there is more than one. There are intervals of every possible duration that all start with the Big Bang. And similar behavior occurs at the other end. So 'primordial interval' is a category of time interval, not an individual. And "all time before x" is the primordial interval that ends with x. Similarly, "in perpetuity after x" is the perpetual interval that starts with x. This is one possible alternative approach. On the other hand, what is needed, according to Issue 16993, is verb concepts to relate situations to indefinite time. Does that require this ontological commitment at all? An alternative approach is to specify just such verbs, e.g.: occurrence is in perpetuity after time interval Definition: the occurrence occurs throughout all time intervals that are after the time interval occurrence is at all times before time interval Definition: the occurrence occurs throughout all time intervals that are before the time interval occurrence is eternal Definition: the occurrence occurs throughout all time intervals I believe this approach was discussed. The problem with this approach is that such occurrences do not have an occurrence interval. On one side or the other there is no time interval that the occurrence does not occur within. That is, these things are "indefinite occurrences". And indefinite occurrences don't have a duration, either. But is that important? Is it a problem that 'all times before time interval1' is not a time interval? Is it important that 'all times before time interval' is a time interval? The underlying question is whether a concept like "the term of a contract" is always a time interval. So I think we are between a rock and a hard place here. One of the following commitments must be made: a) not all time intervals have a duration b) the Time Axis is finite c) not all occurrences (or contracts) have an occurrence interval or a duration I agree with the decision to go with (b). But upon reflection (above), I think the resolution to 17540 needs more than a patch. That is, I think primordial and perpetual are categories of time interval, and 'all time up to x', 'all time thru x', 'all time from x on' (aka 'from x in perpetuity') and 'all time after x', are verbs that characterize primordial and perpetual time intervals. And the binary fact types described above can be defined to relate occurrences to such time intervals. (BTW, I think the additions to the vocabulary in the 16993 proposal are good, but they don't address the issue directly; they only enable the issue to be addressed by using the artificial time intervals.) -Ed -- Edward J. Barkmeyer Email: edbark@nist.gov National Institute of Standards & Technology Manufacturing Systems Integration Division 100 Bureau Drive, Stop 8263 Tel: +1 301-975-3528 Gaithersburg, MD 20899-8263 Cel: +1 240-672-5800 "The opinions expressed above do not reflect consensus of NIST, and have not been reviewed by any Government authority." -- Edward J. Barkmeyer Email: edbark@nist.gov National Institute of Standards & Technology Manufacturing Systems Integration Division 100 Bureau Drive, Stop 8263 Tel: +1 301-975-3528 Gaithersburg, MD 20899-8263 Cel: +1 240-672-5800 "The opinions expressed above do not reflect consensus of NIST, and have not been reviewed by any Government authority." Date: Thu, 20 Sep 2012 11:59:07 -0400 From: Edward Barkmeyer Reply-To: Organization: NIST User-Agent: Thunderbird 2.0.0.23 (Windows/20090812) To: Mark H Linehan CC: OMG DateTimeVoc FTF Subject: Re: DTV Issue 17540: problem with the draft, question about approach Mark H Linehan wrote: Ed, I'm ok with your proposed definition: 'time interval that for each time interval1, either the time interval starts the time interval1 or the time interval starts before the time interval1'. With this definition, I think 'primordiality' is an individual concept. I don't know what I can say that I haven' t already said. It seems to me that there is an infinite set of distinct time intervals, each having a different duration, that satisfy that definition. Each starts with the Big Bang and is distinguished by its duration. Each starts before every time interval that does not start with the Big Bang, and starts with all the ones that do. As I said before, if there is to be a unique one, it would have to have duration 0. Mathematically, it is the limit of the sequence as duration goes to 0, but that limit is not a time interval, exactly as Mike pointed out. Regarding 'January 1,2013 through perpetuity' -- it seems to me that we cannot know whether 'perpetuity' represents a time interval or not, at least not until we reach it. That is why we talk about making an ontological commitment. We don't know that there is one, we just assume that it exists in order to make our model work. By specifying 'perpetuity' as an individual concept, we avoid committing to that choice. This correctly models the world that we experience. We name something so that we can talk about it, even though we don't know now, and may never know, whether the individual concept has an instance. The problem is, however, that if it doesn't correspond to any time interval, then expressions involving it are meaningless. That is 'January 13,2012 through perpetuity' is said to be a "noun form" of 'time interval3 is time interval1 through time interval2", where the role 'time interval1' is played by the referent of 'January 13, 2012' and the role 'time interval2' is played by the referent of 'perpetuity'. If 'perpetuity' has NO referent, then there is no such instance of that verb concept, and there is no time interval3 that satisfies it and is the referent of the "noun form". That is, "The contract between ABC and XYZ holds from 13.1.2012 to perpetuity" is meaningless because the expression "from 13.1.2012 to perpetuity" does not refer to a time interval. The contract holds throughout . The expression is only meaningful if there is a referent of 'perpetuity'. That is why we make the ontological commitment. -Ed ----------------------------- Mark H. Linehan STSM, IBM Research From: Edward Barkmeyer To: Mark H Linehan/Watson/IBM@IBMUS, Cc: OMG DateTimeVoc FTF Date: 09/19/2012 11:53 AM Subject: Re: DTV Issue 17540: problem with the draft, question about approach -------------------------------------------------------------------------------- Mark H Linehan wrote: Ed, I don't understand why my proposed definition of 'primordiality' defines a category rather than an individual concept. I think there is a maximum of one time interval that starts before every other time interval. It seems to me that 'primordiality' starts before any time interval t that is a proper part of 'primordiality'. The problem is exactly what Mike pointed out. The primordiality time interval has to have duration 0. Let it have duration one microsecond, and there is a time inteval that starts it and has duration 1 nanosecond. And primordiality does not start before the 1 nanosecond time interval, and that one does not start before the one that is 1 picosecond in duration, etc. There are no atoms, so the primordiality time interval has a proper part, and if the proper part does not start the primordiality, then there is a time interval that starts the primodiality and meets the proper part. As Mike said, this is all a consequence of the axioms of the interval model -- you can't really model an 'instant'. As defined in clause 8.1.3, 'starts before' subsumes 'starts'. So there is no reason to use "... starts with ... or starts before ...." As defined in 8.1.4, 'starts before' does not subsume 'starts'. Even if it did, it doesn't change the fact that primordiality is a general concept. All of the proper parts that start a primordial time interval are primordial time intervals by the given definition. I think a reference to 'January 1,2013 through perpetuity' need not commit one to the idea that there is a 'perpetuity' time interval. It just commits the time interval to start at the given date and keep going. No. If there is no 'perpetuity' time interval, then the expression 'January 1,2013 through perpetuity' is meaningless, or at best the proposition that includes it is false. If you say "the King of the USA was married on 5 September 2002", what does it mean? It is either meaningless or false, because "the King of the USA" has no referent. If there is a perpetuity then the time interval ends with it. And if there is no perpetuity, then the time interval just continues. If there is no perpetuity, then 'January 1,2013 through perpetuity' does not refer to a time interval. Maybe it will reach 'perpetuity' some day -- but that is not for any of us to know. I have no objection to adding another note about that. The note will say that we are trying to make a silk purse out of a sow's ear and no pig was harmed in constructing it. The proposed model requires the ontological commitment to 'the end of time'. -Ed ----------------------------- Mark H. Linehan STSM, IBM Research From: Edward Barkmeyer To: Mark H Linehan/Watson/IBM@IBMUS, Cc: OMG DateTimeVoc FTF Date: 09/18/2012 06:57 PM Subject: Re: DTV Issue 17540: problem with the draft, question about approach -------------------------------------------------------------------------------- Mark H Linehan wrote: Ed, I think the problem is with our current definition of 'primordiality': "time interval that is before each time interval that is not primordiality". I think it should be "time interval that starts before each time interval that is not primordiality". This latter definition allows 'primordiality' to have proper parts. OK. I like that fix. But with that definition, it does not seem that 'primordiality' is an individual concept. It seems to be a category of time interval, and thus should read "each time interval that is not A primordiality", but that definition might then be considered circular. If t is a primordiality, there is a proper part of t that starts t, and that proper part is then also a primordiality. Note that I suggested the definition: time interval that starts with or before each time interval. But that was ambiguous. What was meant is 'time interval such that for each time interval1, either time interval starts with time interval1 or time interval starts before time interval1. Every time interval that it starts with is also a primordial time interval, by definition. A similar solution using 'ends after' would work for 'perpetuity'. I do not believe we are assuming a finite Time Axis with these definitions. You are right. What I meant by 'finite' is that the Time Axis is itself a time interval that has a fixed duration. And that does not follow. What I was trying to say is that the Time Axis is bounded above and below as a space of time intervals under the ordering 'before'. We don't place these time intervals at the earliest/latest time intervals but without saying that they exist. I don't quite understand. The definitions do make them earliest and latest. You do not intend to assert that the category 'perpetuity' has an instance? If not, then a reference to 'January 1,2013 through perpetuity' doesn't necessarily have a meaning. Nor do we need to make any of the three ontological commitments that you list. I agree that if you don't insist that perpetuity has a referent, then you make no ontological commitment. Presumably, however, anyone who wants to use 'perpetuity' in a contract must make that commitment. Is that the intent? If we are being so careful as that, I think it would be appropriate to add a Note to that effect. Mike Bennett wrote: Yes, my point was that in talking of the time point at the beginning of time we are necessarily thinking in time points. But that these are not time points which decompose into time intervals. I see that you were thinking in terms of time "instants" (time intervals of duration 0). My point was that the concept of perpetuity doesn't require a 'time instant' at the "end of time", or even a "dies irae" (which has duration 1 day?). It can be thought of as "any time interval that ends at the end of time", e.g., "from the signing of the contract forward". Or it can be thought of as "all time intervals after some specified point in time", like the signing of a contract. Apart from Gabriel's trumpet blast, we probably do not need to speak of things that occur "AT the end of time"; we only need to talk about things that occur over intervals that end at the end of time, or occur over all time intervals after some point. -Ed -- Edward J. Barkmeyer Email: edbark@nist.gov National Institute of Standards & Technology Manufacturing Systems Integration Division 100 Bureau Drive, Stop 8263 Tel: +1 301-975-3528 Gaithersburg, MD 20899-8263 Cel: +1 240-672-5800 "The opinions expressed above do not reflect consensus of NIST, and have not been reviewed by any Government authority." ----------------------------- Mark H. Linehan STSM, IBM Research From: Ed Barkmeyer To: Mark H Linehan/Watson/IBM@IBMUS, OMG DateTimeVoc FTF , Date: 09/18/2012 12:54 PM Subject: DTV Issue 17540: problem with the draft, question about approach -------------------------------------------------------------------------------- It occurred to me last night literally in the middle of the night that the idea of primodiality and perpetuity as proposed in 17540 is logically inconsistent with the anti-atomic axiom in some other resolved issue. The axiom says: every time interval has at least one proper part that is a time interval. Now, since perpetuity is said to be a time interval, it follows that it must have a proper part, but it cannot be after any proper part of itself. And therefore there is a time interval that is not perpetuity and is not before perpetuity. One possible fix is to change the definitions to eliminate all time intervals that are part of perpetuity and primordiality. But are these artifice time intervals the best way of describing the finite Time Axis? I think these things are "time instants" (of duration 0) in disguise. That is, they really have duration 0 and no proper parts. If the Time Axis is finite, then there is at least one time interval that is before or starts with every time interval. But there is more than one. There are intervals of every possible duration that all start with the Big Bang. And similar behavior occurs at the other end. So 'primordial interval' is a category of time interval, not an individual. And "all time before x" is the primordial interval that ends with x. Similarly, "in perpetuity after x" is the perpetual interval that starts with x. This is one possible alternative approach. On the other hand, what is needed, according to Issue 16993, is verb concepts to relate situations to indefinite time. Does that require this ontological commitment at all? An alternative approach is to specify just such verbs, e.g.: occurrence is in perpetuity after time interval Definition: the occurrence occurs throughout all time intervals that are after the time interval occurrence is at all times before time interval Definition: the occurrence occurs throughout all time intervals that are before the time interval occurrence is eternal Definition: the occurrence occurs throughout all time intervals I believe this approach was discussed. The problem with this approach is that such occurrences do not have an occurrence interval. On one side or the other there is no time interval that the occurrence does not occur within. That is, these things are "indefinite occurrences". And indefinite occurrences don't have a duration, either. But is that important? Is it a problem that 'all times before time interval1' is not a time interval? Is it important that 'all times before time interval' is a time interval? The underlying question is whether a concept like "the term of a contract" is always a time interval. So I think we are between a rock and a hard place here. One of the following commitments must be made: a) not all time intervals have a duration b) the Time Axis is finite c) not all occurrences (or contracts) have an occurrence interval or a duration I agree with the decision to go with (b). But upon reflection (above), I think the resolution to 17540 needs more than a patch. That is, I think primordial and perpetual are categories of time interval, and 'all time up to x', 'all time thru x', 'all time from x on' (aka 'from x in perpetuity') and 'all time after x', are verbs that characterize primordial and perpetual time intervals. And the binary fact types described above can be defined to relate occurrences to such time intervals. (BTW, I think the additions to the vocabulary in the 16993 proposal are good, but they don't address the issue directly; they only enable the issue to be addressed by using the artificial time intervals.) -Ed -- Edward J. Barkmeyer Email: edbark@nist.gov National Institute of Standards & Technology Manufacturing Systems Integration Division 100 Bureau Drive, Stop 8263 Tel: +1 301-975-3528 Gaithersburg, MD 20899-8263 Cel: +1 240-672-5800 "The opinions expressed above do not reflect consensus of NIST, and have not been reviewed by any Government authority." -- Edward J. Barkmeyer Email: edbark@nist.gov National Institute of Standards & Technology Manufacturing Systems Integration Division 100 Bureau Drive, Stop 8263 Tel: +1 301-975-3528 Gaithersburg, MD 20899-8263 Cel: +1 240-672-5800 "The opinions expressed above do not reflect consensus of NIST, and have not been reviewed by any Government authority." -- Edward J. Barkmeyer Email: edbark@nist.gov National Institute of Standards & Technology Manufacturing Systems Integration Division 100 Bureau Drive, Stop 8263 Tel: +1 301-975-3528 Gaithersburg, MD 20899-8263 Cel: +1 240-672-5800 "The opinions expressed above do not reflect consensus of NIST, and have not been reviewed by any Government authority." Date: Thu, 20 Sep 2012 12:56:08 -0400 From: Ed Barkmeyer Reply-To: Organization: NIST User-Agent: Thunderbird 2.0.0.23 (Windows/20090812) To: "Barkmeyer, Edward J" CC: Mark H Linehan , OMG DateTimeVoc FTF Subject: Re: DTV Issue 17540: problem with the draft, question about approach X-NISTMEL-MailScanner-Information: Please contact postmaster@mel.nist.gov for more information X-NISTMEL-MailScanner-ID: q8KGuFEB016025 X-NISTMEL-MailScanner: Found to be clean X-NISTMEL-MailScanner-SpamCheck: X-NISTMEL-MailScanner-From: edbark@nist.gov X-NISTMEL-MailScanner-Watermark: 1348764977.76418@o8ZpZsfp2rlTsnH/q61x9Q X-Spam-Status: No What I should have mentioned in my response is that "in perpetuity after 13 Jan 2012" is an instance of the category "perpetuity"! It is a time interval that finishes after or with every time interval. But that phrasing would depend on a verb concept like: time interval1 is in perpetuity after time interval2 which means that time interval1 is always an instance of 'perpetuity'. But this makes it necessary to add more verb concepts. The current proposed approach (re)uses time interval3 is time interval1 through time interval2 but that makes it necessary to define a specific 'perpetuity' interval to be time interval2. There is no free lunch. -Ed I wrote: Mark H Linehan wrote: Ed, I'm ok with your proposed definition: 'time interval that for each time interval1, either the time interval starts the time interval1 or the time interval starts before the time interval1'. With this definition, I think 'primordiality' is an individual concept. I don't know what I can say that I haven' t already said. It seems to me that there is an infinite set of distinct time intervals, each having a different duration, that satisfy that definition. Each starts with the Big Bang and is distinguished by its duration. Each starts before every time interval that does not start with the Big Bang, and starts with all the ones that do. As I said before, if there is to be a unique one, it would have to have duration 0. Mathematically, it is the limit of the sequence as duration goes to 0, but that limit is not a time interval, exactly as Mike pointed out. Regarding 'January 1,2013 through perpetuity' -- it seems to me that we cannot know whether 'perpetuity' represents a time interval or not, at least not until we reach it. That is why we talk about making an ontological commitment. We don't know that there is one, we just assume that it exists in order to make our model work. By specifying 'perpetuity' as an individual concept, we avoid committing to that choice. This correctly models the world that we experience. We name something so that we can talk about it, even though we don't know now, and may never know, whether the individual concept has an instance. The problem is, however, that if it doesn't correspond to any time interval, then expressions involving it are meaningless. That is 'January 13,2012 through perpetuity' is said to be a "noun form" of 'time interval3 is time interval1 through time interval2", where the role 'time interval1' is played by the referent of 'January 13, 2012' and the role 'time interval2' is played by the referent of 'perpetuity'. If 'perpetuity' has NO referent, then there is no such instance of that verb concept, and there is no time interval3 that satisfies it and is the referent of the "noun form". That is, "The contract between ABC and XYZ holds from 13.1.2012 to perpetuity" is meaningless because the expression "from 13.1.2012 to perpetuity" does not refer to a time interval. The contract holds throughout . The expression is only meaningful if there is a referent of 'perpetuity'. That is why we make the ontological commitment. -Ed ----------------------------- Mark H. Linehan STSM, IBM Research From: Edward Barkmeyer To: Mark H Linehan/Watson/IBM@IBMUS, Cc: OMG DateTimeVoc FTF Date: 09/19/2012 11:53 AM Subject: Re: DTV Issue 17540: problem with the draft, question about approach -------------------------------------------------------------------------------- Mark H Linehan wrote: Ed, I don't understand why my proposed definition of 'primordiality' defines a category rather than an individual concept. I think there is a maximum of one time interval that starts before every other time interval. It seems to me that 'primordiality' starts before any time interval t that is a proper part of 'primordiality'. The problem is exactly what Mike pointed out. The primordiality time interval has to have duration 0. Let it have duration one microsecond, and there is a time inteval that starts it and has duration 1 nanosecond. And primordiality does not start before the 1 nanosecond time interval, and that one does not start before the one that is 1 picosecond in duration, etc. There are no atoms, so the primordiality time interval has a proper part, and if the proper part does not start the primordiality, then there is a time interval that starts the primodiality and meets the proper part. As Mike said, this is all a consequence of the axioms of the interval model -- you can't really model an 'instant'. As defined in clause 8.1.3, 'starts before' subsumes 'starts'. So there is no reason to use "... starts with ... or starts before ...." As defined in 8.1.4, 'starts before' does not subsume 'starts'. Even if it did, it doesn't change the fact that primordiality is a general concept. All of the proper parts that start a primordial time interval are primordial time intervals by the given definition. I think a reference to 'January 1,2013 through perpetuity' need not commit one to the idea that there is a 'perpetuity' time interval. It just commits the time interval to start at the given date and keep going. No. If there is no 'perpetuity' time interval, then the expression 'January 1,2013 through perpetuity' is meaningless, or at best the proposition that includes it is false. If you say "the King of the USA was married on 5 September 2002", what does it mean? It is either meaningless or false, because "the King of the USA" has no referent. If there is a perpetuity then the time interval ends with it. And if there is no perpetuity, then the time interval just continues. If there is no perpetuity, then 'January 1,2013 through perpetuity' does not refer to a time interval. Maybe it will reach 'perpetuity' some day -- but that is not for any of us to know. I have no objection to adding another note about that. The note will say that we are trying to make a silk purse out of a sow's ear and no pig was harmed in constructing it. The proposed model requires the ontological commitment to 'the end of time'. -Ed ----------------------------- Mark H. Linehan STSM, IBM Research From: Edward Barkmeyer To: Mark H Linehan/Watson/IBM@IBMUS, Cc: OMG DateTimeVoc FTF Date: 09/18/2012 06:57 PM Subject: Re: DTV Issue 17540: problem with the draft, question about approach -------------------------------------------------------------------------------- Mark H Linehan wrote: Ed, I think the problem is with our current definition of 'primordiality': "time interval that is before each time interval that is not primordiality". I think it should be "time interval that starts before each time interval that is not primordiality". This latter definition allows 'primordiality' to have proper parts. OK. I like that fix. But with that definition, it does not seem that 'primordiality' is an individual concept. It seems to be a category of time interval, and thus should read "each time interval that is not A primordiality", but that definition might then be considered circular. If t is a primordiality, there is a proper part of t that starts t, and that proper part is then also a primordiality. Note that I suggested the definition: time interval that starts with or before each time interval. But that was ambiguous. What was meant is 'time interval such that for each time interval1, either time interval starts with time interval1 or time interval starts before time interval1. Every time interval that it starts with is also a primordial time interval, by definition. A similar solution using 'ends after' would work for 'perpetuity'. I do not believe we are assuming a finite Time Axis with these definitions. You are right. What I meant by 'finite' is that the Time Axis is itself a time interval that has a fixed duration. And that does not follow. What I was trying to say is that the Time Axis is bounded above and below as a space of time intervals under the ordering 'before'. We don't place these time intervals at the earliest/latest time intervals but without saying that they exist. I don't quite understand. The definitions do make them earliest and latest. You do not intend to assert that the category 'perpetuity' has an instance? If not, then a reference to 'January 1,2013 through perpetuity' doesn't necessarily have a meaning. Nor do we need to make any of the three ontological commitments that you list. I agree that if you don't insist that perpetuity has a referent, then you make no ontological commitment. Presumably, however, anyone who wants to use 'perpetuity' in a contract must make that commitment. Is that the intent? If we are being so careful as that, I think it would be appropriate to add a Note to that effect. Mike Bennett wrote: Yes, my point was that in talking of the time point at the beginning of time we are necessarily thinking in time points. But that these are not time points which decompose into time intervals. I see that you were thinking in terms of time "instants" (time intervals of duration 0). My point was that the concept of perpetuity doesn't require a 'time instant' at the "end of time", or even a "dies irae" (which has duration 1 day?). It can be thought of as "any time interval that ends at the end of time", e.g., "from the signing of the contract forward". Or it can be thought of as "all time intervals after some specified point in time", like the signing of a contract. Apart from Gabriel's trumpet blast, we probably do not need to speak of things that occur "AT the end of time"; we only need to talk about things that occur over intervals that end at the end of time, or occur over all time intervals after some point. -Ed -- Edward J. Barkmeyer Email: edbark@nist.gov National Institute of Standards & Technology Manufacturing Systems Integration Division 100 Bureau Drive, Stop 8263 Tel: +1 301-975-3528 Gaithersburg, MD 20899-8263 Cel: +1 240-672-5800 "The opinions expressed above do not reflect consensus of NIST, and have not been reviewed by any Government authority." ----------------------------- Mark H. Linehan STSM, IBM Research From: Ed Barkmeyer To: Mark H Linehan/Watson/IBM@IBMUS, OMG DateTimeVoc FTF , Date: 09/18/2012 12:54 PM Subject: DTV Issue 17540: problem with the draft, question about approach -------------------------------------------------------------------------------- It occurred to me last night literally in the middle of the night that the idea of primodiality and perpetuity as proposed in 17540 is logically inconsistent with the anti-atomic axiom in some other resolved issue. The axiom says: every time interval has at least one proper part that is a time interval. Now, since perpetuity is said to be a time interval, it follows that it must have a proper part, but it cannot be after any proper part of itself. And therefore there is a time interval that is not perpetuity and is not before perpetuity. One possible fix is to change the definitions to eliminate all time intervals that are part of perpetuity and primordiality. But are these artifice time intervals the best way of describing the finite Time Axis? I think these things are "time instants" (of duration 0) in disguise. That is, they really have duration 0 and no proper parts. If the Time Axis is finite, then there is at least one time interval that is before or starts with every time interval. But there is more than one. There are intervals of every possible duration that all start with the Big Bang. And similar behavior occurs at the other end. So 'primordial interval' is a category of time interval, not an individual. And "all time before x" is the primordial interval that ends with x. Similarly, "in perpetuity after x" is the perpetual interval that starts with x. This is one possible alternative approach. On the other hand, what is needed, according to Issue 16993, is verb concepts to relate situations to indefinite time. Does that require this ontological commitment at all? An alternative approach is to specify just such verbs, e.g.: occurrence is in perpetuity after time interval Definition: the occurrence occurs throughout all time intervals that are after the time interval occurrence is at all times before time interval Definition: the occurrence occurs throughout all time intervals that are before the time interval occurrence is eternal Definition: the occurrence occurs throughout all time intervals I believe this approach was discussed. The problem with this approach is that such occurrences do not have an occurrence interval. On one side or the other there is no time interval that the occurrence does not occur within. That is, these things are "indefinite occurrences". And indefinite occurrences don't have a duration, either. But is that important? Is it a problem that 'all times before time interval1' is not a time interval? Is it important that 'all times before time interval' is a time interval? The underlying question is whether a concept like "the term of a contract" is always a time interval. So I think we are between a rock and a hard place here. One of the following commitments must be made: a) not all time intervals have a duration b) the Time Axis is finite c) not all occurrences (or contracts) have an occurrence interval or a duration I agree with the decision to go with (b). But upon reflection (above), I think the resolution to 17540 needs more than a patch. That is, I think primordial and perpetual are categories of time interval, and 'all time up to x', 'all time thru x', 'all time from x on' (aka 'from x in perpetuity') and 'all time after x', are verbs that characterize primordial and perpetual time intervals. And the binary fact types described above can be defined to relate occurrences to such time intervals. (BTW, I think the additions to the vocabulary in the 16993 proposal are good, but they don't address the issue directly; they only enable the issue to be addressed by using the artificial time intervals.) -Ed -- Edward J. Barkmeyer Email: edbark@nist.gov National Institute of Standards & Technology Manufacturing Systems Integration Division 100 Bureau Drive, Stop 8263 Tel: +1 301-975-3528 Gaithersburg, MD 20899-8263 Cel: +1 240-672-5800 "The opinions expressed above do not reflect consensus of NIST, and have not been reviewed by any Government authority." -- Edward J. Barkmeyer Email: edbark@nist.gov National Institute of Standards & Technology Manufacturing Systems Integration Division 100 Bureau Drive, Stop 8263 Tel: +1 301-975-3528 Gaithersburg, MD 20899-8263 Cel: +1 240-672-5800 "The opinions expressed above do not reflect consensus of NIST, and have not been reviewed by any Government authority." -- Edward J. Barkmeyer Email: edbark@nist.gov National Institute of Standards & Technology Manufacturing Systems Integration Division 100 Bureau Drive, Stop 8263 Tel: +1 301-975-3528 Gaithersburg, MD 20899-8263 Cel: +1 240-672-5800 "The opinions expressed above do not reflect consensus of NIST, and have not been reviewed by any Government authority." -- Edward J. Barkmeyer Email: edbark@nist.gov National Institute of Standards & Technology Manufacturing Systems Integration Division 100 Bureau Drive, Stop 8263 Tel: +1 301-975-3528 Gaithersburg, MD 20899-8263 Cel: +1 240-672-5800 "The opinions expressed above do not reflect consensus of NIST, and have not been reviewed by any Government authority." To: date-time-ftf@omg.org Subject: Re: DTV Issue 17540: problem with the draft, question about approach X-KeepSent: BDCCE418:F631EC67-85257A83:000CB7DC; type=4; name=$KeepSent X-Mailer: Lotus Notes Release 8.5.3 September 15, 2011 From: Mark H Linehan Date: Sun, 23 Sep 2012 22:26:26 -0400 X-MIMETrack: Serialize by Router on D01MC604/01/M/IBM(Release 8.5.3FP2IF1|July 25, 2012) at 09/23/2012 22:26:31, Serialize complete at 09/23/2012 22:26:31 x-cbid: 12092402-6078-0000-0000-00000FEA09CB OK, here's a slightly different idea: define 'primordiality' as a (category of) time intervals that start before each occurrence. Do the same for 'perpetuity'. Then define verb concept such as "from time interval x on" as you suggested elsewhere. I agree with your comment about 'ontological commitment'. ----------------------------- Mark H. Linehan STSM, IBM Research From: Edward Barkmeyer To: Mark H Linehan/Watson/IBM@IBMUS, Cc: OMG DateTimeVoc FTF Date: 09/20/2012 12:00 PM Subject: Re: DTV Issue 17540: problem with the draft, question about approach -------------------------------------------------------------------------------- Mark H Linehan wrote: Ed, I'm ok with your proposed definition: 'time interval that for each time interval1, either the time interval starts the time interval1 or the time interval starts before the time interval1'. With this definition, I think 'primordiality' is an individual concept. I don't know what I can say that I haven' t already said. It seems to me that there is an infinite set of distinct time intervals, each having a different duration, that satisfy that definition. Each starts with the Big Bang and is distinguished by its duration. Each starts before every time interval that does not start with the Big Bang, and starts with all the ones that do. As I said before, if there is to be a unique one, it would have to have duration 0. Mathematically, it is the limit of the sequence as duration goes to 0, but that limit is not a time interval, exactly as Mike pointed out. Regarding 'January 1,2013 through perpetuity' -- it seems to me that we cannot know whether 'perpetuity' represents a time interval or not, at least not until we reach it. That is why we talk about making an ontological commitment. We don't know that there is one, we just assume that it exists in order to make our model work. By specifying 'perpetuity' as an individual concept, we avoid committing to that choice. This correctly models the world that we experience. We name something so that we can talk about it, even though we don't know now, and may never know, whether the individual concept has an instance. The problem is, however, that if it doesn't correspond to any time interval, then expressions involving it are meaningless. That is 'January 13,2012 through perpetuity' is said to be a "noun form" of 'time interval3 is time interval1 through time interval2", where the role 'time interval1' is played by the referent of 'January 13, 2012' and the role 'time interval2' is played by the referent of 'perpetuity'. If 'perpetuity' has NO referent, then there is no such instance of that verb concept, and there is no time interval3 that satisfies it and is the referent of the "noun form". That is, "The contract between ABC and XYZ holds from 13.1.2012 to perpetuity" is meaningless because the expression "from 13.1.2012 to perpetuity" does not refer to a time interval. The contract holds throughout . The expression is only meaningful if there is a referent of 'perpetuity'. That is why we make the ontological commitment. -Ed ----------------------------- Mark H. Linehan STSM, IBM Research From: Edward Barkmeyer To: Mark H Linehan/Watson/IBM@IBMUS, Cc: OMG DateTimeVoc FTF Date: 09/19/2012 11:53 AM Subject: Re: DTV Issue 17540: problem with the draft, question about approach -------------------------------------------------------------------------------- Mark H Linehan wrote: Ed, I don't understand why my proposed definition of 'primordiality' defines a category rather than an individual concept. I think there is a maximum of one time interval that starts before every other time interval. It seems to me that 'primordiality' starts before any time interval t that is a proper part of 'primordiality'. The problem is exactly what Mike pointed out. The primordiality time interval has to have duration 0. Let it have duration one microsecond, and there is a time inteval that starts it and has duration 1 nanosecond. And primordiality does not start before the 1 nanosecond time interval, and that one does not start before the one that is 1 picosecond in duration, etc. There are no atoms, so the primordiality time interval has a proper part, and if the proper part does not start the primordiality, then there is a time interval that starts the primodiality and meets the proper part. As Mike said, this is all a consequence of the axioms of the interval model -- you can't really model an 'instant'. As defined in clause 8.1.3, 'starts before' subsumes 'starts'. So there is no reason to use "... starts with ... or starts before ...." As defined in 8.1.4, 'starts before' does not subsume 'starts'. Even if it did, it doesn't change the fact that primordiality is a general concept. All of the proper parts that start a primordial time interval are primordial time intervals by the given definition. I think a reference to 'January 1,2013 through perpetuity' need not commit one to the idea that there is a 'perpetuity' time interval. It just commits the time interval to start at the given date and keep going. No. If there is no 'perpetuity' time interval, then the expression 'January 1,2013 through perpetuity' is meaningless, or at best the proposition that includes it is false. If you say "the King of the USA was married on 5 September 2002", what does it mean? It is either meaningless or false, because "the King of the USA" has no referent. If there is a perpetuity then the time interval ends with it. And if there is no perpetuity, then the time interval just continues. If there is no perpetuity, then 'January 1,2013 through perpetuity' does not refer to a time interval. Maybe it will reach 'perpetuity' some day -- but that is not for any of us to know. I have no objection to adding another note about that. The note will say that we are trying to make a silk purse out of a sow's ear and no pig was harmed in constructing it. The proposed model requires the ontological commitment to 'the end of time'. -Ed ----------------------------- Mark H. Linehan STSM, IBM Research From: Edward Barkmeyer To: Mark H Linehan/Watson/IBM@IBMUS, Cc: OMG DateTimeVoc FTF Date: 09/18/2012 06:57 PM Subject: Re: DTV Issue 17540: problem with the draft, question about approach -------------------------------------------------------------------------------- Mark H Linehan wrote: Ed, I think the problem is with our current definition of 'primordiality': "time interval that is before each time interval that is not primordiality". I think it should be "time interval that starts before each time interval that is not primordiality". This latter definition allows 'primordiality' to have proper parts. OK. I like that fix. But with that definition, it does not seem that 'primordiality' is an individual concept. It seems to be a category of time interval, and thus should read "each time interval that is not A primordiality", but that definition might then be considered circular. If t is a primordiality, there is a proper part of t that starts t, and that proper part is then also a primordiality. Note that I suggested the definition: time interval that starts with or before each time interval. But that was ambiguous. What was meant is 'time interval such that for each time interval1, either time interval starts with time interval1 or time interval starts before time interval1. Every time interval that it starts with is also a primordial time interval, by definition. A similar solution using 'ends after' would work for 'perpetuity'. I do not believe we are assuming a finite Time Axis with these definitions. You are right. What I meant by 'finite' is that the Time Axis is itself a time interval that has a fixed duration. And that does not follow. What I was trying to say is that the Time Axis is bounded above and below as a space of time intervals under the ordering 'before'. We don't place these time intervals at the earliest/latest time intervals but without saying that they exist. I don't quite understand. The definitions do make them earliest and latest. You do not intend to assert that the category 'perpetuity' has an instance? If not, then a reference to 'January 1,2013 through perpetuity' doesn't necessarily have a meaning. Nor do we need to make any of the three ontological commitments that you list. I agree that if you don't insist that perpetuity has a referent, then you make no ontological commitment. Presumably, however, anyone who wants to use 'perpetuity' in a contract must make that commitment. Is that the intent? If we are being so careful as that, I think it would be appropriate to add a Note to that effect. Mike Bennett wrote: Yes, my point was that in talking of the time point at the beginning of time we are necessarily thinking in time points. But that these are not time points which decompose into time intervals. I see that you were thinking in terms of time "instants" (time intervals of duration 0). My point was that the concept of perpetuity doesn't require a 'time instant' at the "end of time", or even a "dies irae" (which has duration 1 day?). It can be thought of as "any time interval that ends at the end of time", e.g., "from the signing of the contract forward". Or it can be thought of as "all time intervals after some specified point in time", like the signing of a contract. Apart from Gabriel's trumpet blast, we probably do not need to speak of things that occur "AT the end of time"; we only need to talk about things that occur over intervals that end at the end of time, or occur over all time intervals after some point. -Ed -- Edward J. Barkmeyer Email: edbark@nist.gov National Institute of Standards & Technology Manufacturing Systems Integration Division 100 Bureau Drive, Stop 8263 Tel: +1 301-975-3528 Gaithersburg, MD 20899-8263 Cel: +1 240-672-5800 "The opinions expressed above do not reflect consensus of NIST, and have not been reviewed by any Government authority." ----------------------------- Mark H. Linehan STSM, IBM Research From: Ed Barkmeyer To: Mark H Linehan/Watson/IBM@IBMUS, OMG DateTimeVoc FTF , Date: 09/18/2012 12:54 PM Subject: DTV Issue 17540: problem with the draft, question about approach -------------------------------------------------------------------------------- It occurred to me last night literally in the middle of the night that the idea of primodiality and perpetuity as proposed in 17540 is logically inconsistent with the anti-atomic axiom in some other resolved issue. The axiom says: every time interval has at least one proper part that is a time interval. Now, since perpetuity is said to be a time interval, it follows that it must have a proper part, but it cannot be after any proper part of itself. And therefore there is a time interval that is not perpetuity and is not before perpetuity. One possible fix is to change the definitions to eliminate all time intervals that are part of perpetuity and primordiality. But are these artifice time intervals the best way of describing the finite Time Axis? I think these things are "time instants" (of duration 0) in disguise. That is, they really have duration 0 and no proper parts. If the Time Axis is finite, then there is at least one time interval that is before or starts with every time interval. But there is more than one. There are intervals of every possible duration that all start with the Big Bang. And similar behavior occurs at the other end. So 'primordial interval' is a category of time interval, not an individual. And "all time before x" is the primordial interval that ends with x. Similarly, "in perpetuity after x" is the perpetual interval that starts with x. This is one possible alternative approach. On the other hand, what is needed, according to Issue 16993, is verb concepts to relate situations to indefinite time. Does that require this ontological commitment at all? An alternative approach is to specify just such verbs, e.g.: occurrence is in perpetuity after time interval Definition: the occurrence occurs throughout all time intervals that are after the time interval occurrence is at all times before time interval Definition: the occurrence occurs throughout all time intervals that are before the time interval occurrence is eternal Definition: the occurrence occurs throughout all time intervals I believe this approach was discussed. The problem with this approach is that such occurrences do not have an occurrence interval. On one side or the other there is no time interval that the occurrence does not occur within. That is, these things are "indefinite occurrences". And indefinite occurrences don't have a duration, either. But is that important? Is it a problem that 'all times before time interval1' is not a time interval? Is it important that 'all times before time interval' is a time interval? The underlying question is whether a concept like "the term of a contract" is always a time interval. So I think we are between a rock and a hard place here. One of the following commitments must be made: a) not all time intervals have a duration b) the Time Axis is finite c) not all occurrences (or contracts) have an occurrence interval or a duration I agree with the decision to go with (b). But upon reflection (above), I think the resolution to 17540 needs more than a patch. That is, I think primordial and perpetual are categories of time interval, and 'all time up to x', 'all time thru x', 'all time from x on' (aka 'from x in perpetuity') and 'all time after x', are verbs that characterize primordial and perpetual time intervals. And the binary fact types described above can be defined to relate occurrences to such time intervals. (BTW, I think the additions to the vocabulary in the 16993 proposal are good, but they don't address the issue directly; they only enable the issue to be addressed by using the artificial time intervals.) -Ed -- Edward J. Barkmeyer Email: edbark@nist.gov National Institute of Standards & Technology Manufacturing Systems Integration Division 100 Bureau Drive, Stop 8263 Tel: +1 301-975-3528 Gaithersburg, MD 20899-8263 Cel: +1 240-672-5800 "The opinions expressed above do not reflect consensus of NIST, and have not been reviewed by any Government authority." -- Edward J. Barkmeyer Email: edbark@nist.gov National Institute of Standards & Technology Manufacturing Systems Integration Division 100 Bureau Drive, Stop 8263 Tel: +1 301-975-3528 Gaithersburg, MD 20899-8263 Cel: +1 240-672-5800 "The opinions expressed above do not reflect consensus of NIST, and have not been reviewed by any Government authority." -- Edward J. Barkmeyer Email: edbark@nist.gov National Institute of Standards & Technology Manufacturing Systems Integration Division 100 Bureau Drive, Stop 8263 Tel: +1 301-975-3528 Gaithersburg, MD 20899-8263 Cel: +1 240-672-5800 "The opinions expressed above do not reflect consensus of NIST, and have not been reviewed by any Government authority." To: date-time-ftf@omg.org Subject: Fw: DTV Issue 17540: problem with the draft, question about approach X-KeepSent: D2670E92:54C0F6E1-85257A83:0035625F; type=4; name=$KeepSent X-Mailer: Lotus Notes Release 8.5.3 September 15, 2011 From: Mark H Linehan Date: Mon, 24 Sep 2012 05:45:45 -0400 X-MIMETrack: Serialize by Router on D01MC604/01/M/IBM(Release 8.5.3FP2IF1|July 25, 2012) at 09/24/2012 05:45:46, Serialize complete at 09/24/2012 05:45:46 x-cbid: 12092409-5112-0000-0000-00000CB3CF99 Here's a further development of the idea: define 'primordiality' as the occurrence interval of some occurrence that is before (or starts with) every other occurrence. With this approach, our ontological commitment is that there exists an earliest and latest occurrence. We are not claiming that they "cover" the Time Axis. ----------------------------- Mark H. Linehan STSM, IBM Research ----- Forwarded by Mark H Linehan/Watson/IBM on 09/24/2012 05:43 AM ----- From: Mark H Linehan/Watson/IBM@IBMUS To: date-time-ftf@omg.org, Date: 09/23/2012 10:28 PM Subject: Re: DTV Issue 17540: problem with the draft, question about approach -------------------------------------------------------------------------------- OK, here's a slightly different idea: define 'primordiality' as a (category of) time intervals that start before each occurrence. Do the same for 'perpetuity'. Then define verb concept such as "from time interval x on" as you suggested elsewhere. I agree with your comment about 'ontological commitment'. ----------------------------- Mark H. Linehan STSM, IBM Research From: Edward Barkmeyer To: Mark H Linehan/Watson/IBM@IBMUS, Cc: OMG DateTimeVoc FTF Date: 09/20/2012 12:00 PM Subject: Re: DTV Issue 17540: problem with the draft, question about approach -------------------------------------------------------------------------------- Mark H Linehan wrote: Ed, I'm ok with your proposed definition: 'time interval that for each time interval1, either the time interval starts the time interval1 or the time interval starts before the time interval1'. With this definition, I think 'primordiality' is an individual concept. I don't know what I can say that I haven' t already said. It seems to me that there is an infinite set of distinct time intervals, each having a different duration, that satisfy that definition. Each starts with the Big Bang and is distinguished by its duration. Each starts before every time interval that does not start with the Big Bang, and starts with all the ones that do. As I said before, if there is to be a unique one, it would have to have duration 0. Mathematically, it is the limit of the sequence as duration goes to 0, but that limit is not a time interval, exactly as Mike pointed out. Regarding 'January 1,2013 through perpetuity' -- it seems to me that we cannot know whether 'perpetuity' represents a time interval or not, at least not until we reach it. That is why we talk about making an ontological commitment. We don't know that there is one, we just assume that it exists in order to make our model work. By specifying 'perpetuity' as an individual concept, we avoid committing to that choice. This correctly models the world that we experience. We name something so that we can talk about it, even though we don't know now, and may never know, whether the individual concept has an instance. The problem is, however, that if it doesn't correspond to any time interval, then expressions involving it are meaningless. That is 'January 13,2012 through perpetuity' is said to be a "noun form" of 'time interval3 is time interval1 through time interval2", where the role 'time interval1' is played by the referent of 'January 13, 2012' and the role 'time interval2' is played by the referent of 'perpetuity'. If 'perpetuity' has NO referent, then there is no such instance of that verb concept, and there is no time interval3 that satisfies it and is the referent of the "noun form". That is, "The contract between ABC and XYZ holds from 13.1.2012 to perpetuity" is meaningless because the expression "from 13.1.2012 to perpetuity" does not refer to a time interval. The contract holds throughout . The expression is only meaningful if there is a referent of 'perpetuity'. That is why we make the ontological commitment. -Ed ----------------------------- Mark H. Linehan STSM, IBM Research From: Edward Barkmeyer To: Mark H Linehan/Watson/IBM@IBMUS, Cc: OMG DateTimeVoc FTF Date: 09/19/2012 11:53 AM Subject: Re: DTV Issue 17540: problem with the draft, question about approach -------------------------------------------------------------------------------- Mark H Linehan wrote: Ed, I don't understand why my proposed definition of 'primordiality' defines a category rather than an individual concept. I think there is a maximum of one time interval that starts before every other time interval. It seems to me that 'primordiality' starts before any time interval t that is a proper part of 'primordiality'. The problem is exactly what Mike pointed out. The primordiality time interval has to have duration 0. Let it have duration one microsecond, and there is a time inteval that starts it and has duration 1 nanosecond. And primordiality does not start before the 1 nanosecond time interval, and that one does not start before the one that is 1 picosecond in duration, etc. There are no atoms, so the primordiality time interval has a proper part, and if the proper part does not start the primordiality, then there is a time interval that starts the primodiality and meets the proper part. As Mike said, this is all a consequence of the axioms of the interval model -- you can't really model an 'instant'. As defined in clause 8.1.3, 'starts before' subsumes 'starts'. So there is no reason to use "... starts with ... or starts before ...." As defined in 8.1.4, 'starts before' does not subsume 'starts'. Even if it did, it doesn't change the fact that primordiality is a general concept. All of the proper parts that start a primordial time interval are primordial time intervals by the given definition. I think a reference to 'January 1,2013 through perpetuity' need not commit one to the idea that there is a 'perpetuity' time interval. It just commits the time interval to start at the given date and keep going. No. If there is no 'perpetuity' time interval, then the expression 'January 1,2013 through perpetuity' is meaningless, or at best the proposition that includes it is false. If you say "the King of the USA was married on 5 September 2002", what does it mean? It is either meaningless or false, because "the King of the USA" has no referent. If there is a perpetuity then the time interval ends with it. And if there is no perpetuity, then the time interval just continues. If there is no perpetuity, then 'January 1,2013 through perpetuity' does not refer to a time interval. Maybe it will reach 'perpetuity' some day -- but that is not for any of us to know. I have no objection to adding another note about that. The note will say that we are trying to make a silk purse out of a sow's ear and no pig was harmed in constructing it. The proposed model requires the ontological commitment to 'the end of time'. -Ed ----------------------------- Mark H. Linehan STSM, IBM Research From: Edward Barkmeyer To: Mark H Linehan/Watson/IBM@IBMUS, Cc: OMG DateTimeVoc FTF Date: 09/18/2012 06:57 PM Subject: Re: DTV Issue 17540: problem with the draft, question about approach -------------------------------------------------------------------------------- Mark H Linehan wrote: Ed, I think the problem is with our current definition of 'primordiality': "time interval that is before each time interval that is not primordiality". I think it should be "time interval that starts before each time interval that is not primordiality". This latter definition allows 'primordiality' to have proper parts. OK. I like that fix. But with that definition, it does not seem that 'primordiality' is an individual concept. It seems to be a category of time interval, and thus should read "each time interval that is not A primordiality", but that definition might then be considered circular. If t is a primordiality, there is a proper part of t that starts t, and that proper part is then also a primordiality. Note that I suggested the definition: time interval that starts with or before each time interval. But that was ambiguous. What was meant is 'time interval such that for each time interval1, either time interval starts with time interval1 or time interval starts before time interval1. Every time interval that it starts with is also a primordial time interval, by definition. A similar solution using 'ends after' would work for 'perpetuity'. I do not believe we are assuming a finite Time Axis with these definitions. You are right. What I meant by 'finite' is that the Time Axis is itself a time interval that has a fixed duration. And that does not follow. What I was trying to say is that the Time Axis is bounded above and below as a space of time intervals under the ordering 'before'. We don't place these time intervals at the earliest/latest time intervals but without saying that they exist. I don't quite understand. The definitions do make them earliest and latest. You do not intend to assert that the category 'perpetuity' has an instance? If not, then a reference to 'January 1,2013 through perpetuity' doesn't necessarily have a meaning. Nor do we need to make any of the three ontological commitments that you list. I agree that if you don't insist that perpetuity has a referent, then you make no ontological commitment. Presumably, however, anyone who wants to use 'perpetuity' in a contract must make that commitment. Is that the intent? If we are being so careful as that, I think it would be appropriate to add a Note to that effect. Mike Bennett wrote: Yes, my point was that in talking of the time point at the beginning of time we are necessarily thinking in time points. But that these are not time points which decompose into time intervals. I see that you were thinking in terms of time "instants" (time intervals of duration 0). My point was that the concept of perpetuity doesn't require a 'time instant' at the "end of time", or even a "dies irae" (which has duration 1 day?). It can be thought of as "any time interval that ends at the end of time", e.g., "from the signing of the contract forward". Or it can be thought of as "all time intervals after some specified point in time", like the signing of a contract. Apart from Gabriel's trumpet blast, we probably do not need to speak of things that occur "AT the end of time"; we only need to talk about things that occur over intervals that end at the end of time, or occur over all time intervals after some point. -Ed -- Edward J. Barkmeyer Email: edbark@nist.gov National Institute of Standards & Technology Manufacturing Systems Integration Division 100 Bureau Drive, Stop 8263 Tel: +1 301-975-3528 Gaithersburg, MD 20899-8263 Cel: +1 240-672-5800 "The opinions expressed above do not reflect consensus of NIST, and have not been reviewed by any Government authority." ----------------------------- Mark H. Linehan STSM, IBM Research From: Ed Barkmeyer To: Mark H Linehan/Watson/IBM@IBMUS, OMG DateTimeVoc FTF , Date: 09/18/2012 12:54 PM Subject: DTV Issue 17540: problem with the draft, question about approach -------------------------------------------------------------------------------- It occurred to me last night literally in the middle of the night that the idea of primodiality and perpetuity as proposed in 17540 is logically inconsistent with the anti-atomic axiom in some other resolved issue. The axiom says: every time interval has at least one proper part that is a time interval. Now, since perpetuity is said to be a time interval, it follows that it must have a proper part, but it cannot be after any proper part of itself. And therefore there is a time interval that is not perpetuity and is not before perpetuity. One possible fix is to change the definitions to eliminate all time intervals that are part of perpetuity and primordiality. But are these artifice time intervals the best way of describing the finite Time Axis? I think these things are "time instants" (of duration 0) in disguise. That is, they really have duration 0 and no proper parts. If the Time Axis is finite, then there is at least one time interval that is before or starts with every time interval. But there is more than one. There are intervals of every possible duration that all start with the Big Bang. And similar behavior occurs at the other end. So 'primordial interval' is a category of time interval, not an individual. And "all time before x" is the primordial interval that ends with x. Similarly, "in perpetuity after x" is the perpetual interval that starts with x. This is one possible alternative approach. On the other hand, what is needed, according to Issue 16993, is verb concepts to relate situations to indefinite time. Does that require this ontological commitment at all? An alternative approach is to specify just such verbs, e.g.: occurrence is in perpetuity after time interval Definition: the occurrence occurs throughout all time intervals that are after the time interval occurrence is at all times before time interval Definition: the occurrence occurs throughout all time intervals that are before the time interval occurrence is eternal Definition: the occurrence occurs throughout all time intervals I believe this approach was discussed. The problem with this approach is that such occurrences do not have an occurrence interval. On one side or the other there is no time interval that the occurrence does not occur within. That is, these things are "indefinite occurrences". And indefinite occurrences don't have a duration, either. But is that important? Is it a problem that 'all times before time interval1' is not a time interval? Is it important that 'all times before time interval' is a time interval? The underlying question is whether a concept like "the term of a contract" is always a time interval. So I think we are between a rock and a hard place here. One of the following commitments must be made: a) not all time intervals have a duration b) the Time Axis is finite c) not all occurrences (or contracts) have an occurrence interval or a duration I agree with the decision to go with (b). But upon reflection (above), I think the resolution to 17540 needs more than a patch. That is, I think primordial and perpetual are categories of time interval, and 'all time up to x', 'all time thru x', 'all time from x on' (aka 'from x in perpetuity') and 'all time after x', are verbs that characterize primordial and perpetual time intervals. And the binary fact types described above can be defined to relate occurrences to such time intervals. (BTW, I think the additions to the vocabulary in the 16993 proposal are good, but they don't address the issue directly; they only enable the issue to be addressed by using the artificial time intervals.) -Ed -- Edward J. Barkmeyer Email: edbark@nist.gov National Institute of Standards & Technology Manufacturing Systems Integration Division 100 Bureau Drive, Stop 8263 Tel: +1 301-975-3528 Gaithersburg, MD 20899-8263 Cel: +1 240-672-5800 "The opinions expressed above do not reflect consensus of NIST, and have not been reviewed by any Government authority." -- Edward J. Barkmeyer Email: edbark@nist.gov National Institute of Standards & Technology Manufacturing Systems Integration Division 100 Bureau Drive, Stop 8263 Tel: +1 301-975-3528 Gaithersburg, MD 20899-8263 Cel: +1 240-672-5800 "The opinions expressed above do not reflect consensus of NIST, and have not been reviewed by any Government authority." -- Edward J. Barkmeyer Email: edbark@nist.gov National Institute of Standards & Technology Manufacturing Systems Integration Division 100 Bureau Drive, Stop 8263 Tel: +1 301-975-3528 Gaithersburg, MD 20899-8263 Cel: +1 240-672-5800 "The opinions expressed above do not reflect consensus of NIST, and have not been reviewed by any Government authority." To: date-time-ftf@omg.org Subject: DTV issue 17540 - Need to support infinite and indefinite time constructs X-KeepSent: FF6A50E4:19335183-85257A88:007288F0; type=4; name=$KeepSent X-Mailer: Lotus Notes Release 8.5.3 September 15, 2011 From: Mark H Linehan Date: Sat, 29 Sep 2012 16:54:53 -0400 X-MIMETrack: Serialize by Router on D01MC604/01/M/IBM(Release 8.5.3FP2IF1|July 25, 2012) at 09/29/2012 16:54:55 x-cbid: 12092920-3534-0000-0000-00000D446460 Here is a new attempt on this issue, based on the idea that 'primordiality' is the time interval of the earliest occurrence. ----------------------------- Mark H. Linehan STSM, IBM Research Date-Time Issue 17540 - Need to Support Infinite and Indefinite Time Constructs2.doc Disposition: Resolved OMG Issue No: 17540 Title: Need to support infinite and indefinite time constructs Source: Mike Bennett, EDM Council, mbennett@edmcouncil.org Summary: The Date Time Vocabulary needs to be able to support statements about periods of time which extend indefinitely into the future, and also describe periods of time which will have begun at indeterminate times in the past. As an example of the former, it is possible and meaningful for a contract to make statements about commitments or rights which extend in perpetuity, such as Perpetual Bonds which are bonds that pay interest forever. In general, it is necessary to be able to make meaningful statements which embody the concept of Forever. Similarly, it is necessary to be able to make meaningful statements which have been going on in perpetuity up to the present time. The DTV specification does currently allow for making statements about infinite time going forward, but not about time periods which have started at some indefinite time in the past. Meanwhile there are three other issues in play which touch on this same area. These may have an impact on the current ability to make statements about infinite time into the future as well, depending on their final resolution. The issues which have a bearing or potential bearing on this matter are: - Issue 16992: Corollaries to Axiom D.4 in 8.2.3 are misstated; - Issue 16993: no syntax for indefinite time periods (date-time-ftf); - Issue 16997: forever is misdefined In summary, the requirement that needs to be met with the resolution of this and the above-referenced issues is: 1. Extend forever with two new concepts: a. Indeterminate time in the past b. Indefinite time in the future 2. Ensure that the concept forever can be adequately defined (per 16997) including with reference to the time axis; 3. That there is syntax for the specification of indeterminate time periods that began at some point in the past and last up until the present (per 16993) 4. That the restatement of the axiom and corollaries referenced in 16992 take account of the two concepts above (indeterminate time in the past and indefinite time in the future) Resolution: The FTF decided that there is no value in distinguishing .indefinite. from .infinite.. It chose to add new concepts that provide the basis for time intervals that extend indefinitely into the past or the future. New individual concepts .primordiality. and .perpetuity. are defined respectively as the occurrence interval of the earliest occurrence and of the latest occurrence. .Eternity. (synonym .forever.) is defined as .primordiality through perpetuity.. This permits formulations such as .primordiality through today. and .2012 through perpetuity.. A tool might support these formulations with a syntax such as .. until today .. and .. from 2012 on ... Issue 16993 adds verb concepts such as .time interval until situation model.. .primordiality. and .perpetuity. can substitute for the .time interval. role to enable formulations such as .primordiality until the Industrial Revolution.. Revised Text: (These update instructions apply to the beta-2 specification). Replace figure 8.5 to add two new verb concepts. <> Add new glossary entries to the end of clause 8.1.4 'Additional Time Interval Relationships' time interval1 begins before time interval2 Definition: time interval1 is before time interval2 or time interval1 properly overlaps time interval2 or time interval1 starts time interval2 Description: the start of time interval1 is before the start of time interval2 CLIF Definition: (forall ((ti1 ti2)) (iff ("time interval1 begins before time interval2" ti1 ti2) (and ("time interval" ti1) ("time interval" ti2) (or ("time interval1 is before time interval2" ti1 ti2) ("time interval1 properly overlaps time interval2" ti1 ti2) ("time interval1 starts time interval2" ti1 ti2))))) OCL Definition: context "time interval" def: "time interval1 begins before time interval2" (ti2: "time interval") : Boolean = self."is before"(ti2) or self.overlaps(ti2) or self.starts(ti2) Example: 2012 January 1 begins before 2012 week 1 time interval1 ends after time interval2 Definition: time interval1 is after time interval2 or time interval1 is properly overlapped by time interval2 or time interval1 ends time interval2 Description: the end of time interval1 is after the end of time interval2 CLIF Definition: (forall ((ti1 ti2)) (iff ("time interval1 ends after time interval2" ti1 ti2) (and ("time interval" ti1) ("time interval" ti2) (or ("time interval1 is after time interval2" ti1 ti2) ("time interval1 properly overlaps time interval2" ti2 ti1) ("time interval1 ends time interval2" ti1 ti2))))) OCL Definition: context "time interval" def: "time interval1 ends after time interval2" (ti2: "time interval") : Boolean = self."is after"(ti2) or ti2."properly overlaps"(self) or self.ends(ti2) Example: December ends after December 25 Add a new clause 8.1.8 at the end of clause 8.1. The new clause adds a figure, so the following figures of clause 8 must be renumbered. 8.1.8 Indefinite Time Intervals Indefinite time intervals provide the basis for describing time intervals that extend indefinitely into the past or the future. One example is a British bond of the 1910.s that pays interest .in perpetuity.. <> primordiality Definition: occurrence interval of some occurrence1 that begins before each occurrence2 Description: The time interval of the first occurrence. Note: This concept can be used in formulations such as .primordiality through today. to define time intervals that began at some indefinite time in the past. Tools may choose to support a convenient syntax such as .until today.. Note: 'Primordiality' has a duration but it is not known. Note: 'Primordiality' is an individual concept because there can be only one occurrence that begins before every occurrence. Example: .Primordiality through Tuesday. meaning .until Tuesday.. perpetuity Definition: occurrence interval of some occurrence1 that ends after each occurrence2 Description: The time interval of the latest occurrence. Note: This concept can be used in formulations such as .2012 through perpetuity. to define time intervals that extend indefinitely into the future. Tools may choose to support a convenient syntax such as .after 2012.. Note: 'Perpetuity' has a duration but it is not known. Note: 'Perpetuity' is an individual concept because there can be only one such occurrence. Example: .Contract signing through perpetuity. meaning .after the contract signing.. eternity Synonym: forever Definition: primordiality through perpetuity Description: The time interval that extends across the entire Time Axis. Note: 'Eternity' is an individual concept because there can be only one such time interval. Note: Eternity is not the same thing as the Time Axis, even though it .covers. the Time Axis. In clause 8.6, replace figure 8.19 with this version, which deletes the instance .forever.. <> In clause 8.6, delete the glossary entry for .forever.. Disposition: Resolved To: date-time-ftf@omg.org Subject: Re: DTV issue 17540 - Need to support infinite and indefinite time constructs X-KeepSent: 102028BD:C55322D6-85257A8A:005FD478; type=4; name=$KeepSent X-Mailer: Lotus Notes Release 8.5.3 September 15, 2011 From: Mark H Linehan Date: Mon, 1 Oct 2012 13:29:07 -0400 X-MIMETrack: Serialize by Router on D01MC604/01/M/IBM(Release 8.5.3FP2IF1|July 25, 2012) at 10/01/2012 13:29:09 x-cbid: 12100117-3534-0000-0000-00000D4FE0EB This is a minor update that adds Necessities that say that a time interval can begin before/end after itself. (The verb symbol implies otherwise, but I can't think of a better verb symbol.) ----------------------------- Mark H. Linehan STSM, IBM Research From: Mark H Linehan/Watson/IBM@IBMUS To: date-time-ftf@omg.org, Date: 09/29/2012 04:56 PM Subject: DTV issue 17540 - Need to support infinite and indefinite time constructs -------------------------------------------------------------------------------- Here is a new attempt on this issue, based on the idea that 'primordiality' is the time interval of the earliest occurrence. ----------------------------- Mark H. Linehan STSM, IBM Research[attachment "Date-Time Issue 17540 - Need to Support Infinite and Indefinite Time Constructs.doc" deleted by Mark H Linehan/Watson/IBM] Date-Time Issue 17540 - Need to Support Infinite and Indefinite Time Constructs3.doc Disposition: Resolved OMG Issue No: 17540 Title: Need to support infinite and indefinite time constructs Source: Mike Bennett, EDM Council, mbennett@edmcouncil.org Summary: The Date Time Vocabulary needs to be able to support statements about periods of time which extend indefinitely into the future, and also describe periods of time which will have begun at indeterminate times in the past. As an example of the former, it is possible and meaningful for a contract to make statements about commitments or rights which extend in perpetuity, such as Perpetual Bonds which are bonds that pay interest forever. In general, it is necessary to be able to make meaningful statements which embody the concept of Forever. Similarly, it is necessary to be able to make meaningful statements which have been going on in perpetuity up to the present time. The DTV specification does currently allow for making statements about infinite time going forward, but not about time periods which have started at some indefinite time in the past. Meanwhile there are three other issues in play which touch on this same area. These may have an impact on the current ability to make statements about infinite time into the future as well, depending on their final resolution. The issues which have a bearing or potential bearing on this matter are: - Issue 16992: Corollaries to Axiom D.4 in 8.2.3 are misstated; - Issue 16993: no syntax for indefinite time periods (date-time-ftf); - Issue 16997: forever is misdefined In summary, the requirement that needs to be met with the resolution of this and the above-referenced issues is: 1. Extend forever with two new concepts: a. Indeterminate time in the past b. Indefinite time in the future 2. Ensure that the concept forever can be adequately defined (per 16997) including with reference to the time axis; 3. That there is syntax for the specification of indeterminate time periods that began at some point in the past and last up until the present (per 16993) 4. That the restatement of the axiom and corollaries referenced in 16992 take account of the two concepts above (indeterminate time in the past and indefinite time in the future) Resolution: The FTF decided that there is no value in distinguishing .indefinite. from .infinite.. It chose to add new concepts that provide the basis for time intervals that extend indefinitely into the past or the future. New individual concepts .primordiality. and .perpetuity. are defined respectively as the occurrence interval of the earliest occurrence and of the latest occurrence. .Eternity. (synonym .forever.) is defined as .primordiality through perpetuity.. This permits formulations such as .primordiality through today. and .2012 through perpetuity.. A tool might support these formulations with a syntax such as .. until today .. and .. from 2012 on ... Issue 16993 adds verb concepts such as .time interval until situation model.. .primordiality. and .perpetuity. can substitute for the .time interval. role to enable formulations such as .primordiality until the Industrial Revolution.. Revised Text: (These update instructions apply to the beta-2 specification). Replace figure 8.5 to add two new verb concepts. <> Add new glossary entries to the end of clause 8.1.4 'Additional Time Interval Relationships' time interval1 begins before time interval2 Definition: time interval1 is before time interval2 or time interval1 properly overlaps time interval2 or time interval1 starts time interval2 Description: the start of time interval1 is before the start of time interval2 CLIF Definition: (forall ((ti1 ti2)) (iff ("time interval1 begins before time interval2" ti1 ti2) (and ("time interval" ti1) ("time interval" ti2) (or ("time interval1 is before time interval2" ti1 ti2) ("time interval1 properly overlaps time interval2" ti1 ti2) ("time interval1 starts time interval2" ti1 ti2))))) OCL Definition context "time interval" def: "time interval1 begins before time interval2" (ti2: "time interval") : Boolean = self."is before"(ti2) or self.overlaps(ti2) or self.starts(ti2) Necessity: Each time interval begins before the time interval. CLIF Axiom: (forall (ti) (if ("time interval" ti) ("time interval1 begins before time interval2" ti ti))) OCL Constraint: context "time interval" inv: self."time interval1 begins before time interval2"(self) Example: 2012 January 1 begins before 2012 week 1 time interval1 ends after time interval2 Definition: time interval1 is after time interval2 or time interval1 is properly overlapped by time interval2 or time interval1 ends time interval2 Description: the end of time interval1 is after the end of time interval2 CLIF Definition: (forall ((ti1 ti2)) (iff ("time interval1 ends after time interval2" ti1 ti2) (and ("time interval" ti1) ("time interval" ti2) (or ("time interval1 is after time interval2" ti1 ti2) ("time interval1 properly overlaps time interval2" ti2 ti1) ("time interval1 ends time interval2" ti1 ti2))))) OCL Definition context "time interval" def: "time interval1 ends after time interval2" (ti2: "time interval") : Boolean = self."is after"(ti2) or ti2."properly overlaps"(self) or self.ends(ti2) Necessity: Each time interval ends after the time interval. CLIF Axiom: (forall (ti) (if ("time interval" ti) ("time interval1 ends after time interval2" ti ti))) OCL Constraint: context "time interval" inv: self."time interval1 ends after time interval2"(self) Example: December ends after December 25 Add a new clause 8.1.8 at the end of clause 8.1. The new clause adds a figure, so the following figures of clause 8 must be renumbered. 8.1.8 Indefinite Time Intervals Indefinite time intervals provide the basis for describing time intervals that extend indefinitely into the past or the future. One example is a British bond of the 1910.s that pays interest .in perpetuity.. <> primordiality Definition: occurrence interval of some occurrence1 that begins before each occurrence2 Description: The time interval of the first occurrence. Note: This concept can be used in formulations such as .primordiality through today. to define time intervals that began at some indefinite time in the past. Tools may choose to support a convenient syntax such as .until today.. Note: 'Primordiality' has a duration but it is not known. Note: 'Primordiality' is an individual concept because there can be only one occurrence that begins before every occurrence. Example: .Primordiality through Tuesday. meaning .until Tuesday.. perpetuity Definition: occurrence interval of some occurrence1 that ends after each occurrence2 Description: The time interval of the latest occurrence. Note: This concept can be used in formulations such as .2012 through perpetuity. to define time intervals that extend indefinitely into the future. Tools may choose to support a convenient syntax such as .after 2012.. Note: 'Perpetuity' has a duration but it is not known. Note: 'Perpetuity' is an individual concept because there can be only one such occurrence. Example: .Contract signing through perpetuity. meaning .after the contract signing.. eternity Synonym: forever Definition: primordiality through perpetuity Description: The time interval that extends across the entire Time Axis. Note: 'Eternity' is an individual concept because there can be only one such time interval. Note: Eternity is not the same thing as the Time Axis, even though it .covers. the Time Axis. In clause 8.6, replace figure 8.19 with this version, which deletes the instance .forever.. <> In clause 8.6, delete the glossary entry for .forever.. Disposition: Resolved