Issue 1962: Transmission of type codes across 2.2/2.3 boundaries (interop) Source: (, ) Nature: Uncategorized Issue Severity: Summary: Summary: Suppose I am using a 2.3 ORB and receive a type code from a 2.2 ORB without a repository ID. What am I supposed to do if I want to send that type code somewhere else? (This could happen for an event channel, for example.) Resolution: In this case, we need to add the case for empty repository ID string implies that a repositoryID is Revised Text: page 15-23, second sentence of first paragraph of "Encoded Identifiers and Names" immediatly preceeding Table 15-7: Change sentence: " For GIOP 1.2 onwards, repositoryID values are mandatory. " to the following sentence with footnote: " For GIOP 1.2 onwards, repositoryID values are required to be sent, if known by the ORB. An empty repositoryID string is only allowed if a repositoryID value is not available to the ORB sending the type code. * " <* indicates footnote number> <text of footnote > " * A type code passed via a GIOP 1.2 connection shall contain non-empty repositoryID strings, unless a repositoryID value is not available to the sending ORB for a specific type code. This situation can arise,for example, if an ORB receives a type code containing empty repository IDs via a GIOP 1.0 or 1.1 connection and passes that type code on via a GIOP 1.2 connection). " Actions taken: September 16, 1998: received issue October 25, 1999: moved from core to interop rtf October 4, 2000: closed issue Discussion: This could also happen with run-time specification of property types in property lists, where new properties could be defined by an application, without reference to an IDL module containg the property Type definition. End of Annotations:===== Return-Path: X-Authentication-Warning: fatcat.dstc.edu.au: michi owned process doing -bs Date: Wed, 16 Sep 1998 18:38:29 +1000 (EST) From: Michi Henning To: orb_revision@omg.org Subject: Transmission of type codes across 2.2/2.3 boundaries Suppose I am using a 2.3 ORB and receive a type code from a 2.2 ORB without a repository ID. What am I supposed to do if I want to send that type code somewhere else? (This could happen for an event channel, for example.) Given that my ORB is 2.3 compliant and that repository IDs are mandatory in CORBA 2.3, I could assume that it would be illegal to send a type code to someone else if the type code does not have a repository ID. Clearly, that is not the desired behavior though. I think the spec could do with some words to say so. Secondary, more tricky, question: if I receive a type code that contains a type name and member names from somewhere, am I allowed as a CORBA 2.3 ORB to pass that type code on with its names removed? I suspect we should add a few words to the spec to state that a CORBA 2.3 ORB must preserve all parts of a type code it receives without stripping anything. Otherwise, if we have a 2.3 event channel between a 2.2 supplier and a 2.2 consumer, the type codes may arrive at the consumer in a shape where they have lost information that was originally present, which could break existing client code. Cheers, Michi. -- Michi Henning +61 7 33654310 DSTC Pty Ltd +61 7 33654311 (fax) University of Qld 4072 michi@dstc.edu.au AUSTRALIA http://www.dstc.edu.au/BDU/staff/michi-henning.html To: interop@omg.org Subject: Are repository IDs mandatory in GIOP 1.2 typecodes? From: Bill Janssen Message-Id: <00May8.172552pdt."3438"@watson.parc.xerox.com> Date: Mon, 8 May 2000 17:26:04 PDT Content-Type: text X-UIDL: M4o!!DgJe9!6Z!!/Z8e9 In the current interop vote, the proposed resolution to 1962 seems to exactly reverse the sense of the section. That is, it changes repository IDs in typecodes from being mandatory to being optional. The proposed text includes the statement: ``Note that a type code passed via a GIOP 1.2 connection may contain empty repository ID values, even though the repository ID parameters are mandatory for GIOP 1.2.''. This is nonsense; we should either remove the idea that repository IDs are mandatory, or (my preference) state that typecodes received with repository IDs cannot be passed on to a GIOP 1.2 connection. I urge you all to vote no on 1962. Bill Date: Tue, 9 May 2000 10:58:31 +1000 (EST) From: Michi Henning To: Bill Janssen cc: interop@omg.org Subject: Re: Are repository IDs mandatory in GIOP 1.2 typecodes? In-Reply-To: <00May8.172552pdt."3438"@watson.parc.xerox.com> Message-ID: Organization: Object Oriented Concepts MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII X-UIDL: ]@Je9+p;e9k)!"!ifpd9 On Mon, 8 May 2000, Bill Janssen wrote: > In the current interop vote, the proposed resolution to 1962 seems to > exactly reverse the sense of the section. That is, it changes > repository IDs in typecodes from being mandatory to being optional. > > The proposed text includes the statement: ``Note that a type code > passed via a GIOP 1.2 connection may contain empty repository ID > values, even though the repository ID parameters are mandatory for GIOP > 1.2.''. This is nonsense; we should either remove the idea that > repository IDs are mandatory, or (my preference) state that typecodes > received with repository IDs cannot be passed on to a GIOP 1.2 > connection. > > I urge you all to vote no on 1962. Bill, if we disallow a 1.2 implementation to carry type codes without repository IDs, 1.1 and 1.2 can't interoperate anymore. Where would, say, a receiving 1.2 ORB take the repository IDs from if the came from a 1.1 ORB? Cheers, Michi. -- Michi Henning +61 7 3891 5744 Object Oriented Concepts +61 4 1118 2700 (mobile) Suite 4, 904 Stanley St +61 7 3891 5009 (fax) East Brisbane 4169 michi@ooc.com.au AUSTRALIA http://www.ooc.com.au/staff/michi-henning.html Sender: jon@corvette.floorboard.com Message-ID: <39176967.7D5DBF58@floorboard.com> Date: Mon, 08 May 2000 18:27:03 -0700 From: Jonathan Biggar X-Mailer: Mozilla 4.7 [en] (X11; U; SunOS 5.5.1 sun4m) X-Accept-Language: en MIME-Version: 1.0 To: Bill Janssen CC: interop@omg.org Subject: Re: Are repository IDs mandatory in GIOP 1.2 typecodes? References: <00May8.172552pdt."3438"@watson.parc.xerox.com> Content-Transfer-Encoding: 7bit Content-Type: text/plain; charset=us-ascii X-UIDL: \f`!!"B6e9N(3e9HR6e9 Bill Janssen wrote: > > In the current interop vote, the proposed resolution to 1962 seems to > exactly reverse the sense of the section. That is, it changes > repository IDs in typecodes from being mandatory to being optional. > > The proposed text includes the statement: ``Note that a type code > passed via a GIOP 1.2 connection may contain empty repository ID > values, even though the repository ID parameters are mandatory for GIOP > 1.2.''. This is nonsense; we should either remove the idea that > repository IDs are mandatory, or (my preference) state that typecodes > received with repository IDs cannot be passed on to a GIOP 1.2 > connection. The intent was to allow TypeCodes without repository id information to be received from GIOP 1.1- ORBs and be passed on legally. Your preference breaks interoperability if an intermediary service, like an Event server is upgraded to GIOP 1.2 but some clients remain GIOP 1.1. How about this: Change: "For GIOP 1.2 onwards, all TypeCodes that originate in an ORB must be encoded with non-null repositoryId values. ORBs must accept incoming TypeCodes without repositoryId values from a GIOP 1.2 connection without signalling an error." to: and change the footnote to: "Since an ORB that receives a TypeCode that does not contain a repository ID via a GIOP 1.0 or 1.1 connection must be able to retransmit it without the repository ID on a GIOP 1.2 connection, " -- Jon Biggar Floorboard Software jon@floorboard.com jon@biggar.org To: Michi Henning cc: interop@omg.org Subject: Re: Are repository IDs mandatory in GIOP 1.2 typecodes? In-Reply-To: Your message of "Mon, 08 May 2000 17:58:51 PDT." From: Bill Janssen Message-Id: <00May8.184406pdt."3438"@watson.parc.xerox.com> Date: Mon, 8 May 2000 18:44:16 PDT Content-Type: text X-UIDL: 7)cd9_mPe9ZUY!!NpBe9 > Bill, if we disallow a 1.2 implementation to carry type codes without > repository IDs, 1.1 and 1.2 can't interoperate anymore. Where would, say, > a receiving 1.2 ORB take the repository IDs from if the came from a 1.1 ORB? I'm afraid it would have to reject them. Or, actually, not send them on; it could presumably still use them internally for something. If you feel that interoperability between (1.0, 1.1) and 1.2 is more important, then I suggest we instead remove all that nonsense about repository IDs being mandatory. You can have one or the other, but not both. Bill Date: Tue, 9 May 2000 12:21:24 +1000 (EST) From: Michi Henning To: Bill Janssen cc: interop@omg.org Subject: Re: Are repository IDs mandatory in GIOP 1.2 typecodes? In-Reply-To: <00May8.184406pdt."3438"@watson.parc.xerox.com> Message-ID: Organization: Object Oriented Concepts MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII X-UIDL: O#?e9E)F!!k5,e9nK1!! On Mon, 8 May 2000, Bill Janssen wrote: > > Bill, if we disallow a 1.2 implementation to carry type codes without > > repository IDs, 1.1 and 1.2 can't interoperate anymore. Where would, say, > > a receiving 1.2 ORB take the repository IDs from if the came from a 1.1 ORB? > > I'm afraid it would have to reject them. Or, actually, not send them > on; it could presumably still use them internally for something. > > If you feel that interoperability between (1.0, 1.1) and 1.2 is more > important, then I suggest we instead remove all that nonsense about > repository IDs being mandatory. You can have one or the other, but > not both. Hmmm... Repository IDs were made mandatory because, otherwise, we can't have proper type checking. In particular, the difference between TypeCode::equal and TypeCode::equivalent is crucial. (I believe that most people would agree that making repository IDs optional again would be a retrograde step.) So, the only remaining options I can see are: 1) have a 1.2 ORB reject all type codes that don't have repository IDs where 1.2 requires them 2) have a 1.2 ORB allow such type codes to come in but don't allow a 1.2 ORB to ever create such type codes The consequence of option 1 is that interoperability between 1.1 and 1.2 ORBs goes out the window as soon as an Any is involved. The consequence of option 2 is that interoperability is still possible, but that operations like TypeCode::equal and TypeCode::equivalent have strange outcomes if 1.1 type codes are involved. I prefer option 2, and option 2 is what we decided on when we added TypeCode::equivalent. Cheers, Michi. -- Michi Henning +61 7 3891 5744 Object Oriented Concepts +61 4 1118 2700 (mobile) Suite 4, 904 Stanley St +61 7 3891 5009 (fax) East Brisbane 4169 michi@ooc.com.au AUSTRALIA http://www.ooc.com.au/staff/michi-henning.html Date: Tue, 9 May 2000 12:26:03 +1000 (EST) From: Michi Henning To: Jonathan Biggar cc: Bill Janssen , interop@omg.org Subject: Re: Are repository IDs mandatory in GIOP 1.2 typecodes? In-Reply-To: <39176967.7D5DBF58@floorboard.com> Message-ID: Organization: Object Oriented Concepts MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII X-UIDL: M)j!!LGYd9[hed9\T,e9 On Mon, 8 May 2000, Jonathan Biggar wrote: > How about this: > > Change: > > "For GIOP 1.2 onwards, all TypeCodes that originate in an ORB must > be > encoded with non-null repositoryId values. ORBs must accept > incoming > TypeCodes without repositoryId values from a GIOP 1.2 connection > without > signalling an error." > > to: > > and change the footnote to: > > "Since an ORB that receives a TypeCode that does not contain a > repository ID via a GIOP 1.0 or 1.1 connection must be able to > retransmit it without the repository ID on a GIOP 1.2 connection, " Sounds OK to me. Cheers, Michi. -- Michi Henning +61 7 3891 5744 Object Oriented Concepts +61 4 1118 2700 (mobile) Suite 4, 904 Stanley St +61 7 3891 5009 (fax) East Brisbane 4169 michi@ooc.com.au AUSTRALIA http://www.ooc.com.au/staff/michi-henning.html From: Paul Kyzivat To: interop@omg.org Subject: RE: Are repository IDs mandatory in GIOP 1.2 typecodes? Date: Tue, 9 May 2000 08:27:14 -0400 MIME-Version: 1.0 X-Mailer: Internet Mail Service (5.5.2448.0) Content-Type: text/plain; charset="iso-8859-1" X-UIDL: E\)!!1Ie9 I agree with Michi and Jon. I understand Bill's point that this weakens the type system, but backward compatibility is too important to give up. This at least makes progress - the type system will have the improved strength when all players are 1.2 or beyond. Paul > -----Original Message----- > From: Bill Janssen [mailto:janssen@parc.xerox.com] > Sent: Monday, May 08, 2000 8:26 PM > To: interop@omg.org > Subject: Are repository IDs mandatory in GIOP 1.2 typecodes? > > > In the current interop vote, the proposed resolution to 1962 seems to > exactly reverse the sense of the section. That is, it changes > repository IDs in typecodes from being mandatory to being optional. > > The proposed text includes the statement: ``Note that a type code > passed via a GIOP 1.2 connection may contain empty repository ID > values, even though the repository ID parameters are > mandatory for GIOP > 1.2.''. This is nonsense; we should either remove the idea that > repository IDs are mandatory, or (my preference) state that typecodes > received with repository IDs cannot be passed on to a GIOP 1.2 > connection. > > I urge you all to vote no on 1962. > > Bill > From: "Rutt, T E (Tom)" To: Bill Janssen , "'Jonathan Biggar'" Cc: interop@omg.org Subject: RE: Are repository IDs mandatory in GIOP 1.2 typecodes? Date: Tue, 9 May 2000 11:01:05 -0400 MIME-Version: 1.0 X-Mailer: Internet Mail Service (5.5.2448.0) Content-Type: text/plain X-UIDL: SF6!!D2jd9=co!!3/$!! If we are wanting to be really "Strong Minded" we could require a situation where the mediator orb has to use multiple GIOP versions on the outgoing connection to handle this situation. If the repId is not present on a 1.1 giop message, then use giop 1.1 to "relay" it. Is this alternative "better" than explaining that "missing means unavailable". ? Either way, the receiving orb does not get the repositoryId. -- Tom Rutt Lucent Technologies - Bell Labs Rm 4L-336 Tel: +1(732)949-7862 101 Crawford Corner Rd Fax: +1(732)949-1196 Holmdel NJ, 07733 USA email: terutt@lucent.com > ---------- > From: Jonathan Biggar[SMTP:jon@floorboard.com] > Sent: Monday, May 08, 2000 9:27 PM > To: Bill Janssen > Cc: interop@omg.org > Subject: Re: Are repository IDs mandatory in GIOP 1.2 typecodes? > > Bill Janssen wrote: > > > > In the current interop vote, the proposed resolution to 1962 seems to > > exactly reverse the sense of the section. That is, it changes > > repository IDs in typecodes from being mandatory to being optional. > > > > The proposed text includes the statement: ``Note that a type code > > passed via a GIOP 1.2 connection may contain empty repository ID > > values, even though the repository ID parameters are mandatory for GIOP > > 1.2.''. This is nonsense; we should either remove the idea that > > repository IDs are mandatory, or (my preference) state that typecodes > > received with repository IDs cannot be passed on to a GIOP 1.2 > > connection. > > The intent was to allow TypeCodes without repository id information to > be received from GIOP 1.1- ORBs and be passed on legally. Your > preference breaks interoperability if an intermediary service, like an > Event server is upgraded to GIOP 1.2 but some clients remain GIOP 1.1. > > How about this: > > Change: > > "For GIOP 1.2 onwards, all TypeCodes that originate in an ORB must be > encoded with non-null repositoryId values. ORBs must accept incoming > TypeCodes without repositoryId values from a GIOP 1.2 connection without > signalling an error." > > to: > > and change the footnote to: > > "Since an ORB that receives a TypeCode that does not contain a > repository ID via a GIOP 1.0 or 1.1 connection must be able to > retransmit it without the repository ID on a GIOP 1.2 connection, " > > -- > Jon Biggar > Floorboard Software > jon@floorboard.com > jon@biggar.org > From: Paul Kyzivat To: interop@omg.org Subject: RE: Are repository IDs mandatory in GIOP 1.2 typecodes? Date: Tue, 9 May 2000 13:42:00 -0400 MIME-Version: 1.0 X-Mailer: Internet Mail Service (5.5.2448.0) Content-Type: text/plain; charset="iso-8859-1" X-UIDL: Y]L!!D8U!!&7K!!BmWd9 Tom - sorry, but this doesn't always work. Suppose I receive a GIOP 1.2 request message, and then discover that my reply will contain an Any that contains a typecode without a required repository id? I have no choice but to reply using GIOP 1.2. Paul > -----Original Message----- > From: Rutt, T E (Tom) [mailto:terutt@lucent.com] > Sent: Tuesday, May 09, 2000 11:01 AM > To: Bill Janssen; 'Jonathan Biggar' > Cc: interop@omg.org > Subject: RE: Are repository IDs mandatory in GIOP 1.2 typecodes? > > > If we are wanting to be really "Strong Minded" we could > require a situation > where the > mediator orb has to use multiple GIOP versions on the > outgoing connection > to > handle this situation. If the repId is not present on a 1.1 > giop message, > then use > giop 1.1 to "relay" it. > > Is this alternative "better" than explaining that "missing means > unavailable". ? > > Either way, the receiving orb does not get the repositoryId. > > > -- > Tom Rutt > Lucent Technologies - Bell Labs > Rm 4L-336 Tel: +1(732)949-7862 > 101 Crawford Corner Rd Fax: +1(732)949-1196 > Holmdel NJ, 07733 USA email: terutt@lucent.com > > > > > ---------- > > From: Jonathan Biggar[SMTP:jon@floorboard.com] > > Sent: Monday, May 08, 2000 9:27 PM > > To: Bill Janssen > > Cc: interop@omg.org > > Subject: Re: Are repository IDs mandatory in GIOP 1.2 typecodes? > > > > Bill Janssen wrote: > > > > > > In the current interop vote, the proposed resolution to > 1962 seems to > > > exactly reverse the sense of the section. That is, it changes > > > repository IDs in typecodes from being mandatory to being > optional. > > > > > > The proposed text includes the statement: ``Note that a type code > > > passed via a GIOP 1.2 connection may contain empty repository ID > > > values, even though the repository ID parameters are > mandatory for GIOP > > > 1.2.''. This is nonsense; we should either remove the idea that > > > repository IDs are mandatory, or (my preference) state > that typecodes > > > received with repository IDs cannot be passed on to a GIOP 1.2 > > > connection. > > > > The intent was to allow TypeCodes without repository id > information to > > be received from GIOP 1.1- ORBs and be passed on legally. Your > > preference breaks interoperability if an intermediary > service, like an > > Event server is upgraded to GIOP 1.2 but some clients > remain GIOP 1.1. > > > > How about this: > > > > Change: > > > > "For GIOP 1.2 onwards, all TypeCodes that originate in an > ORB must be > > encoded with non-null repositoryId values. ORBs must > accept incoming > > TypeCodes without repositoryId values from a GIOP 1.2 > connection without > > signalling an error." > > > > to: > > > > and change the footnote to: > > > > "Since an ORB that receives a TypeCode that does not contain a > > repository ID via a GIOP 1.0 or 1.1 connection must be able to > > retransmit it without the repository ID on a GIOP 1.2 connection, " > > > > -- > > Jon Biggar > > Floorboard Software > > jon@floorboard.com > > jon@biggar.org > > > Date: Wed, 10 May 2000 08:58:00 +1000 (EST) From: Michi Henning To: "Rutt, T E (Tom)" cc: Bill Janssen , "'Jonathan Biggar'" , interop@omg.org Subject: RE: Are repository IDs mandatory in GIOP 1.2 typecodes? In-Reply-To: Message-ID: Organization: Object Oriented Concepts MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII X-UIDL: VJ;!!f_$!!B]'e9W7Y!! On Tue, 9 May 2000, Rutt, T E (Tom) wrote: > If we are wanting to be really "Strong Minded" we could require a situation > where the > mediator orb has to use multiple GIOP versions on the outgoing connection > to > handle this situation. If the repId is not present on a 1.1 giop message, > then use > giop 1.1 to "relay" it. > > Is this alternative "better" than explaining that "missing means > unavailable". ? Hmmm... I don't think I like this idea. It would require a lot of checking in the code and make the already complex connection multiplexing logic even more complex. Besides, I don't think making this change would gain anything, because it would still permit a 1.2 ORB to receive an Any without repository IDs. Whether they are delivered over 1.2 or 1.1 doesn't really matter, does it? My feeling is that it would be best to leave things as they are. Cheers, Michi. -- Michi Henning +61 7 3891 5744 Object Oriented Concepts +61 4 1118 2700 (mobile) Suite 4, 904 Stanley St +61 7 3891 5009 (fax) East Brisbane 4169 michi@ooc.com.au AUSTRALIA http://www.ooc.com.au/staff/michi-henning.html To: Michi Henning cc: interop@omg.org Subject: Re: Are repository IDs mandatory in GIOP 1.2 typecodes? In-Reply-To: Your message of "Mon, 08 May 2000 19:21:34 PDT." From: Bill Janssen Message-Id: <00May9.161606pdt."3438"@watson.parc.xerox.com> Date: Tue, 9 May 2000 16:16:12 PDT Content-Type: text X-UIDL: B+-!!;BT!!5`J!!SB&e9 > So, the only remaining options I can see are: > > 1) have a 1.2 ORB reject all type codes that don't have > repository IDs > where 1.2 requires them > > 2) have a 1.2 ORB allow such type codes to come in but > don't allow > a 1.2 ORB to ever create such type codes > > The consequence of option 1 is that interoperability between 1.1 and > 1.2 ORBs goes out the window as soon as an Any is involved. > > The consequence of option 2 is that interoperability is still > possible, > but that operations like TypeCode::equal and TypeCode::equivalent > have > strange outcomes if 1.1 type codes are involved. > > I prefer option 2, and option 2 is what we decided on when we added > TypeCode::equivalent. I guess the question is, is the ORB allowed to send such typecodes on to another ORB over a GIOP 1.2 connection? If so, we should remove any wording in the interop section about such repository IDs being mandatory, because they clearly aren't -- any receiver has to be able to cope with typecodes that don't have them, and any sender can *claim* to have received those typecodes from elsewhere. The best we can do is to strongly suggest that typecodes include the repository ID. Personally, I'd be in favor of option 1. But hey! That's just me :-). Bill Sender: jbiggar@corvette.floorboard.com Message-ID: <3918A412.17719E9C@floorboard.com> Date: Tue, 09 May 2000 16:49:38 -0700 From: Jonathan Biggar X-Mailer: Mozilla 4.7 [en] (X11; U; SunOS 5.6 sun4u) X-Accept-Language: en MIME-Version: 1.0 To: Bill Janssen CC: Michi Henning , interop@omg.org Subject: Re: Are repository IDs mandatory in GIOP 1.2 typecodes? References: <00May9.161606pdt."3438"@watson.parc.xerox.com> Content-Transfer-Encoding: 7bit Content-Type: text/plain; charset=us-ascii X-UIDL: 1Vdd9;<>e9k<&!!E*N!! Bill Janssen wrote: > > So, the only remaining options I can see are: > > > > 1) have a 1.2 ORB reject all type codes that don't have repository IDs > > where 1.2 requires them > > > > 2) have a 1.2 ORB allow such type codes to come in but don't allow > > a 1.2 ORB to ever create such type codes > > > > The consequence of option 1 is that interoperability between 1.1 and > > 1.2 ORBs goes out the window as soon as an Any is involved. > > > > The consequence of option 2 is that interoperability is still possible, > > but that operations like TypeCode::equal and TypeCode::equivalent have > > strange outcomes if 1.1 type codes are involved. > > > > I prefer option 2, and option 2 is what we decided on when we added > > TypeCode::equivalent. > > I guess the question is, is the ORB allowed to send such typecodes on > to another ORB over a GIOP 1.2 connection? If so, we should remove > any wording in the interop section about such repository IDs being > mandatory, because they clearly aren't -- any receiver has to be able > to cope with typecodes that don't have them, and any sender can > *claim* to have received those typecodes from elsewhere. The best we > can do is to strongly suggest that typecodes include the repository > ID. What we need, as I had in my counter-proposal was a requirement that ORBs supporting GIOP 1.2 that originate a typecode must include the repository ids, but it is still acceptable to retransmit a received typecode that does not have repository ids. > Personally, I'd be in favor of option 1. But hey! That's just me :-). You better watch out, or someone will claim you are a Micro$oft sleeper agent. :-) -- Jon Biggar Floorboard Software jon@floorboard.com jon@biggar.org Date: Wed, 10 May 2000 10:00:02 +1000 (EST) From: Michi Henning To: Bill Janssen cc: interop@omg.org Subject: Re: Are repository IDs mandatory in GIOP 1.2 typecodes? In-Reply-To: <00May9.161606pdt."3438"@watson.parc.xerox.com> Message-ID: Organization: Object Oriented Concepts MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII X-UIDL: O*%!!Y&'!!3k$!!+H^d9 On Tue, 9 May 2000, Bill Janssen wrote: > > I prefer option 2, and option 2 is what we decided on when we added > > TypeCode::equivalent. > > I guess the question is, is the ORB allowed to send such typecodes on > to another ORB over a GIOP 1.2 connection? Yes, otherwise we lose interoperability. > If so, we should remove > any wording in the interop section about such repository IDs being > mandatory, because they clearly aren't -- any receiver has to be > able > to cope with typecodes that don't have them, and any sender can > *claim* to have received those typecodes from elsewhere. Yes. > The best we > can do is to strongly suggest that typecodes include the repository > ID. We might be able to do better. We could require a 1.2 ORB to never produce a type code without the repository ID (which is what the current proposal does, I think). Can you suggest stronger words that would capture this better? > Personally, I'd be in favor of option 1. But hey! That's just me :-). Exactly :-) Cheers, Michi. -- Michi Henning +61 7 3891 5744 Object Oriented Concepts +61 4 1118 2700 (mobile) Suite 4, 904 Stanley St +61 7 3891 5009 (fax) East Brisbane 4169 michi@ooc.com.au AUSTRALIA http://www.ooc.com.au/staff/michi-henning.html To: Paul Kyzivat cc: interop@omg.org Subject: Re: Are repository IDs mandatory in GIOP 1.2 typecodes? In-Reply-To: Your message of "Tue, 09 May 2000 05:25:07 PDT." <9B164B713EE9D211B6DC0090273CEEA926BE7C@bos1.noblenet.com> From: Bill Janssen Message-Id: <00May9.182124pdt."3438"@watson.parc.xerox.com> Date: Tue, 9 May 2000 18:21:34 PDT Content-Type: text X-UIDL: jC>!!DZn!!8_1e9R)[d9 Paul, I'm not suggesting we give up on backward compatibility. I'm suggesting that if we decide to have backward compatibility, all the verbiage about mandatory typecode repository IDs is nonsense. An ORB can not count on ever receiving a repository ID in a typecode. This makes the proposed resolution gibberish. Vote No, and let's fix it in the next go-round. Bill > I agree with Michi and Jon. > I understand Bill's point that this weakens the type system, but > backward > compatibility is too important to give up. This at least makes > progress - > the type system will have the improved strength when all players are > 1.2 or > beyond. > > Paul > > > -----Original Message----- > > From: Bill Janssen [mailto:janssen@parc.xerox.com] > > Sent: Monday, May 08, 2000 8:26 PM > > To: interop@omg.org > > Subject: Are repository IDs mandatory in GIOP 1.2 typecodes? > > > > > > In the current interop vote, the proposed resolution to 1962 seems > to > > exactly reverse the sense of the section. That is, it changes > > repository IDs in typecodes from being mandatory to being > optional. > > > > The proposed text includes the statement: ``Note that a type code > > passed via a GIOP 1.2 connection may contain empty repository ID > > values, even though the repository ID parameters are > > mandatory for GIOP > > 1.2.''. This is nonsense; we should either remove the idea that > > repository IDs are mandatory, or (my preference) state that > typecodes > > received with repository IDs cannot be passed on to a GIOP 1.2 > > connection. > > > > I urge you all to vote no on 1962. > > > > Bill > > To: Michi Henning cc: interop@omg.org Subject: Re: Are repository IDs mandatory in GIOP 1.2 typecodes? In-Reply-To: Your message of "Tue, 09 May 2000 17:00:14 PDT." From: Bill Janssen Message-Id: <00May9.182629pdt."3438"@watson.parc.xerox.com> Date: Tue, 9 May 2000 18:26:33 PDT Content-Type: text X-UIDL: 6,Ae9H,l!!~PJe9"DL!! > We might be able to do better. We could require a 1.2 ORB to never produce > a type code without the repository ID (which is what the current proposal > does, I think). Can you suggest stronger words that would capture this > better? I like the general idea. But I think you're talking apples and oranges; it's not the interop part of the ORB which produces the typecode -- GIOP is just concerned with what gets put on the wire. What I hear everyone saying is that you can put a typecode on the wire either with or without the repository ID, to allow for backwards compatibility. If either is valid, an ORB has no specific reason to bend over backwards to produce one if it's inconvenient. Bill Date: Wed, 10 May 2000 12:20:45 +1000 (EST) From: Michi Henning To: Bill Janssen cc: interop@omg.org Subject: Re: Are repository IDs mandatory in GIOP 1.2 typecodes? In-Reply-To: <00May9.182629pdt."3438"@watson.parc.xerox.com> Message-ID: Organization: Object Oriented Concepts MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII X-UIDL: CPM!!go)!!)gB!!"TQ!! On Tue, 9 May 2000, Bill Janssen wrote: > > We might be able to do better. We could require a 1.2 ORB to never produce > > a type code without the repository ID (which is what the current proposal > > does, I think). Can you suggest stronger words that would capture this > > better? > > I like the general idea. But I think you're talking apples and > oranges; it's not the interop part of the ORB which produces the > typecode -- GIOP is just concerned with what gets put on the wire. > What I hear everyone saying is that you can put a typecode on the wire > either with or without the repository ID, to allow for backwards > compatibility. If either is valid, an ORB has no specific reason > to bend over backwards to produce one if it's inconvenient. Yes, formally, you are right. Could we say something in the core spec about this? Somewhere near the TypeCode interface definition with an XREF to the GIOP chapter? Cheers, Michi. -- Michi Henning +61 7 3891 5744 Object Oriented Concepts +61 4 1118 2700 (mobile) Suite 4, 904 Stanley St +61 7 3891 5009 (fax) East Brisbane 4169 michi@ooc.com.au AUSTRALIA http://www.ooc.com.au/staff/michi-henning.html Date: Wed, 10 May 2000 12:24:29 +1000 (EST) From: Michi Henning To: Bill Janssen cc: Paul Kyzivat , interop@omg.org Subject: Re: Are repository IDs mandatory in GIOP 1.2 typecodes? In-Reply-To: <00May9.182124pdt."3438"@watson.parc.xerox.com> Message-ID: Organization: Object Oriented Concepts MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII X-UIDL: U5+e9^EPd9,4g!!65De9 On Tue, 9 May 2000, Bill Janssen wrote: > Paul, > > I'm not suggesting we give up on backward compatibility. I'm > suggesting that if we decide to have backward compatibility, all the > verbiage about mandatory typecode repository IDs is nonsense. An > ORB > can not count on ever receiving a repository ID in a typecode. This > makes the proposed resolution gibberish. Vote No, and let's fix it > in > the next go-round. I'm beginning to come round to your view. The requirement really must be attached to CORBA 2.4, not to GIOP 1.2. The problem is that the encoding of type codes is described in chapter 15... Cheers, Michi. Date: Wed, 10 May 2000 12:24:29 +1000 (EST) From: Michi Henning To: Bill Janssen cc: Paul Kyzivat , interop@omg.org Subject: Re: Are repository IDs mandatory in GIOP 1.2 typecodes? In-Reply-To: <00May9.182124pdt."3438"@watson.parc.xerox.com> Message-ID: Organization: Object Oriented Concepts MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII X-UIDL: U5+e9^EPd9,4g!!65De9 On Tue, 9 May 2000, Bill Janssen wrote: > Paul, > > I'm not suggesting we give up on backward compatibility. I'm > suggesting that if we decide to have backward compatibility, all the > verbiage about mandatory typecode repository IDs is nonsense. An > ORB > can not count on ever receiving a repository ID in a typecode. This > makes the proposed resolution gibberish. Vote No, and let's fix it > in > the next go-round. I'm beginning to come round to your view. The requirement really must be attached to CORBA 2.4, not to GIOP 1.2. The problem is that the encoding of type codes is described in chapter 15... Cheers, Michi. To: Michi Henning cc: interop@omg.org Subject: Re: Are repository IDs mandatory in GIOP 1.2 typecodes? In-Reply-To: Your message of "Tue, 09 May 2000 17:00:14 PDT." From: Bill Janssen Message-Id: <00May10.155616pdt."3438"@watson.parc.xerox.com> Date: Wed, 10 May 2000 15:56:20 PDT Content-Type: text X-UIDL: b1md9PhW!!/+Ud9m]4e9 > > Personally, I'd be in favor of option 1. But hey! That's just me :-). > > Exactly :-) Or perhaps, "Usually" :-). Bill From: "Rutt, T E (Tom)" To: Bill Janssen , "'Michi Henning'" Cc: interop@omg.org Subject: RE: Are repository IDs mandatory in GIOP 1.2 typecodes? Date: Thu, 11 May 2000 14:14:44 -0400 MIME-Version: 1.0 X-Mailer: Internet Mail Service (5.5.2448.0) Content-Type: text/plain X-UIDL: NRe!!"?m!!P2g!!BRTd9 The issue is complicated by the DII. If a user can make an any value using the DII (say perhaps for a property list), and that user does not supply repository IDs over the DII api, What is the orb supposed to do when sending that property list in an operation parameter using GIOP 1.2? If the DII allows the user to construct any values with typecodes not having repository ID values, then the wording "if available" is appropriate for the GIOP 1.2 requirement for typecodes. -- Tom Rutt Lucent Technologies - Bell Labs Rm 4L-336 Tel: +1(732)949-7862 101 Crawford Corner Rd Fax: +1(732)949-1196 Holmdel NJ, 07733 USA email: terutt@lucent.com > ---------- > From: Michi Henning[SMTP:michi@ooc.com.au] > Sent: Tuesday, May 09, 2000 10:20 PM > To: Bill Janssen > Cc: interop@omg.org > Subject: Re: Are repository IDs mandatory in GIOP 1.2 typecodes? > > On Tue, 9 May 2000, Bill Janssen wrote: > > > > We might be able to do better. We could require a 1.2 ORB to never > produce > > > a type code without the repository ID (which is what the current > proposal > > > does, I think). Can you suggest stronger words that would capture this > > > better? > > > > I like the general idea. But I think you're talking apples and > > oranges; it's not the interop part of the ORB which produces the > > typecode -- GIOP is just concerned with what gets put on the wire. > > What I hear everyone saying is that you can put a typecode on the wire > > either with or without the repository ID, to allow for backwards > > compatibility. If either is valid, an ORB has no specific reason > > to bend over backwards to produce one if it's inconvenient. > > Yes, formally, you are right. Could we say something in the core spec > about this? Somewhere near the TypeCode interface definition with an XREF > to the GIOP chapter? > > Cheers, > > Michi. > -- > Michi Henning +61 7 3891 5744 > Object Oriented Concepts +61 4 1118 2700 (mobile) > Suite 4, 904 Stanley St +61 7 3891 5009 (fax) > East Brisbane 4169 michi@ooc.com.au > AUSTRALIA http://www.ooc.com.au/staff/michi-henning.html > Sender: jbiggar@corvette.floorboard.com Message-ID: <391B4A27.F6F4223A@floorboard.com> Date: Thu, 11 May 2000 17:02:47 -0700 From: Jonathan Biggar X-Mailer: Mozilla 4.7 [en] (X11; U; SunOS 5.6 sun4u) X-Accept-Language: en MIME-Version: 1.0 To: interop@omg.org Subject: Re: Are repository IDs mandatory in GIOP 1.2 typecodes? References: Content-Transfer-Encoding: 7bit Content-Type: text/plain; charset=us-ascii X-UIDL: F&Qd9-)Rd9FC-e9YG0e9 "Rutt, T E (Tom)" wrote: > > The issue is complicated by the DII. > > If a user can make an any value using the DII (say perhaps for a property > list), > and that user does not supply repository IDs over the DII api, What is the > orb supposed to do when sending that property list in an operation parameter > using GIOP 1.2? > > If the DII allows the user to construct any values with typecodes not having > repository ID values, then the wording "if available" is appropriate for the > GIOP 1.2 requirement for typecodes. Tom, check the resolution of 2907 from the Core RTF. We voted that the TypeCode construction routines must check that repository IDs are supplied and that they are supplied in the correct syntax. This will apply to CORBA 2.4. -- Jon Biggar Floorboard Software jon@floorboard.com jon@biggar.org