COMMON FACILITIES TASK FORCE Meeting Minutes San Jose, California, June 28 & 29, 1995 Task Force Chair: Tom Mowbray, PhD, MITRE OMG Meeting Host: Ken McCoy, Tandem Minutes recorded and edited by Geoff Speare, OMG and Tom Mowbray, MITRE Attendee List: Manimo Actis-Dato DEC Richard Adler CCT Art Altman EPRI Tat Anthias IBM Dipak Bara Taligent Richard Barnwell Software 2000 Dennis Barrett Tandem Claude Baudoin Schumberger Charles Brauer Fujitsu Tim Brinson Protocol Systems Michael Brodie GTE Dwight Brown DoD George Brown Intel Marina Buchman Informix Alex Buchmann THD Carol Burt Bellsouth Cory Casanave Data Access Steven Case Computing Devices Wes Chadwick Ratheon Athena Chang Protocol Systems Dan Chang IBM Tom Clark Tandem Jean Claude Derniame CRIN Johan Fantenberg Ellemtel Jon Farmer Care Data Systems Liz Farrell Fujitsu Herm Fischer MARKV Linda Fisher Pitney Bowes Elizabeth Fong NIST Dage Gamble Micro Focus Karl Gardner Raytheon Vic Goddard XOpen Prashant Gupta Sybase Dan Harkey IBM Tadashi Hirose Hitachi Richard Hoffman IBM Glenn Holowell Sematech Sridhar Iyengar Unisys Neil Katin CI Labs Mansour Kavianvour Object Enterprise Group Barry Kitson Telstra/TINA-C Rainer Kossman BNR Bob Kostes Platinum Wilfried Kruse Tandem Huet Landry DISA Laurent LeBoucher France Telecom Herve Lejeune Group Bull Mark Linton Silicon Graphics Bil Litman Tandem Denise Lynch UTC Robert MacGregor USC/ISI Heather McKelvey Tandem Mike Meier IBM Nicholas Mercouroff Alcatel/TINA-C Jeff Mischkinsky Sybase Naohiko Mori NTT/NCALS Tom Mowbray MITRE David Newman Technium Dave Oliver BNR Damian Osisek IBM Rudolf Reiss DEC Masoto Saito Hitachi Chris Schoppa HP Ron Skelton EPRI Greg Smith Expersoft Mike Sobolewski GE CRD Jacob Stein Sybase James Storer TCSI Jeff Sutherland VMARK Shel Sutton MITRE Craig Thompson Texas Instruments Steve Tockey Boeing Borek Vokach-Brodsky Apple Mike Zonis Tandem Wednesday, June 28, 1995 Meeting started at 10:04. Tom Mowbray said that I'd like to give a little bit of background for those that are new here. Key CF documents: CF Architecture 95-1-2 CF Roadmap 95-1-32 RFP1 94-9-13 Compound Documents RFI2 95-1-24 Financial Services RFP2 95-1-31 Internationalization & Time Facilities Initial submissions to RFP1 Fresco 95-3-1 OpenDoc 95-3-24 Tom Mowbray went over the agenda for the meeting. Tom Mowbray said that the deadline for revised submissions for RFP1 is 25 October, prior to the Tokyo meeting. Tom Mowbray presented the current roadmap: Start RFP Items Planned adoption Oct94 RFP1 Compound document Oct95-Feb96 Mar95 RFP2 Internationalization & Time Operations Mar96 Sept95 RFP3 Data interchange & Automation Sept96 Sept95 RFP4 Scripting agent facility, Rule Management Sept96 Feb96 RFP5 Information Exchange Feb 96 Roadmap facilities for Sept 95 RFPs: Data interchange (linked to information exchange) Automation Scripting Rule Management Martin Anderson talked about the work of the BOMSIG in this area. We need to start to address things like the work on the Internet that was discussed at the TC this morning. What I'd like to propose is a new working group, with a charter: To form a CF working group to address those horizontal facilities which are of particular relevance to business & application interoperability. Initially: Data Interchange, Information Exchange. (Co-chairs: Dan Chang (IBM), Martin Anderson (IO)). Martin Anderson so moved. Easel (now BMARK) seconded. Craig Thompson asked how this WG relates to the sequencing of the RFPs. Tom Mowbray said that the facilities identified here are on the roadmap (Data Interchange and Information Exchange), and scheduled for release. Martin Anderson said that the purpose of the WG is to drive those two items through with attention to the application of business object issues, and also to be a vehicle to pick up those items in subsequent places on the roadmap where they occur. Don Belisle asked if that means preparing the RFP. Martin Anderson said yes -- this would be the group of people who did the grunt work. Jeff Mischkinsky asked about the relation between this and the appdev working group. Martin Anderson said that each has their own particular focus, though they may need to work together. Jeff Mischkinsky asked what the difference in focus is. Martin Anderson said that the focus for this WG would be on interoperability. Tom Mowbray said that they would be working on different facilities -- appdev is working on repositories originally, and perhaps later workflow. That's quite different than the proposed WG. Someone asked what other areas might come up which might attract the interest of this WG. Martin Anderson said that some of the facilities on the roadmap at present have a very direct impact on the business side of things. I don't want to get too specific, because I don't have a list of the next 3 or 4 facilities. I would like to go through this process to demonstrate that it does add value, and then go from there. Bob Hodges said that one of the appdev's main tasks, an RFI, will be focused on capturing metadata, and exchanging that data. There's probably a need to make sure that the work of appdev is integrated with this WG. Sankar Virdhagriswaran said that not limiting this WG to just two facilities is a good thing. Martin Anderson said that's true, but I also don't want to lay claim yet to any future facilities. Denise Lynch said that the TC asked the BOMSIG to come present on what we are working on; I'd like to see a presentation on that architecture. Tom Mowbray said maybe it would be better to be rescheduled. Rudolf Riess said why do we have to build another separate group to handle this work? Why can't it be handled in the TF. I don't see any real advantage in pushing it down another level. Someone asked if there is a connection between this group and the newly proposed high-level Architecture Committee. Martin Anderson said that there will need to be continual interchange of ideas between groups--that is where the Architecture Committee will be important. Don Belisle said that it seems like overkill to have a working group here. Jishnu Mukerji said that based on the work of the appdev WG, creating another WG would only serve to slow things down. Tom Mowbray said that this WG would be able to do a great deal of work in parallel, and help to declutter the task force agenda. We're asking more than just the simple task item of writing an RFP. Doing this work in parallel will help work go quicker and with more focus. Someone said that the mission statement of the group is rather subjective--the group is given a life of its own. How do you tell what facilities fall under this group? I think it would be better to have a number of transient WGs with specific missions rather than one with a vague and confusing mission statement. Martin Anderson said that coming from this group, we have a number of different strands of technology, which impact the business user >from different perspectives. What I'm proposing is that we have a place so that if there are particular business issues, they have a place to arrive--although they might go elsewhere. If we don't have that place, we lose a link between the two groups. Herm Fischer said that the appdev group has a lot of different things we are working on, we have gone off in a lot of different directions. Jishnu Mukerji said could we not incorporate business concerns as part of the RFP evaluation? Don Belisle agreed--if the business community would like to participate, they should join in the TF discussions. Martin Anderson said that there are a lot of things in here that are not of interest to the business community--what I would like to do is provide an area where they can focus on their concerns. That will bring in their participation. Jeff Mischkinsky said that there is a lot of overlap here -- I don't think a new group is needed. There was discussion. Sankar Virdhagriswaran proposed as an amendment that the mission of this group is driving the RFP process through the Data Interchange and Information Exchange only. Martin Anderson did not accept as a friendly amendment. Tandem seconded. Jeff Mischkinsky said that looking at the roadmap, it seems like these two items are of drastically different priority. Would we be changing that ordering? Sankar Virdhagriswaran said that the two are closely related--we may want to work on both at the same time, regardless of sequencing. Someone said there was discussion about looking at objects on the internet in a working group-- could there be a link there? There was discussion on the mission of the WG were this amendment to pass. Tom Mowbray said that my understanding is that the WG would be responsible for writing an RFP for these two facilities. Sankar Virdhagriswaran said that I think whatever the sequencing, this WG should define the RFP material for these two facilities. The discussion of whether they ought to go together or not is a separate issue. Martin Anderson said that I think we need to get a clear distinction between the work that needs to go in to deliver an RFP, and the question of how we actually assure that all the things the TF are approving are useful for the community. If this work does not include the community of potential users, it may not deliver what we need to deliver as a part of OMG. We need to make this infrastructure stuff mean something to the community. I think it would be useful to hear from users what would be helpful. There are a lot of comments outside this room from users -- it does not serve anyone to keep that gap in place. Don Belisle said that it seems to me the way to do that is to get a clear definition of data interchange, and to steer creation of the RFP and the adoption process. Once submissions are received, the only way to influence what is adopted is to get involved in the voting process. I don't see any other way. Rudolf Riess said that the CF TF has thus far been very responsive to outside interests -- I don't think another WG would help drive the issues. Someone said that we want to make sure to close the gap between the business community. Perhaps a better idea would be to have this WG review and comment on all RFPs, to help influence the voting process. Michael Brodie said that within my company, we've been trying to use OMG technology on a massive scale. One of our big concerns is the view that is presented to the programmer--to make sure that it is as simple as possible. My understanding of this proposal is that it is a broad one--that the two services mentioned are only examples. Speaking as an end-user, I do not want to require our programmers to have to understand every last detail of how everything works. I want to see this issue pursued vigorously. Don Belisle said that it sounds like there's a tendency toward a more formal standards organization, where we can fine tune things and make small changes. OMG doesn't work that way--if you want to work within the OMG process, you need to get out and talk to the submitters, and influence the submissions as they are made. Michael Brodie said that what I'm concerned about is how all of these facilities work together, rather than running from facilities group to group. Sankar Virdhagriswaran said that we are not discussing the merits of getting end-users involved in the process -- what we're talking about is the goal of this particular group. I'm saying that for practical purposes I am proposing we limit the scope of this particular group. I don't think anyone is disagreeing with what you just said. Perhaps there should be a separate motion to do whatever people feel is necessary to make end-user concerns heard. Craig Thompson said that we are hearing over and over again that there are many different concerns that need to be heard. Martin Anderson said that the BOMSIG has discussed this; the best way they feel they can contribute to the process is by speaking within something where there is an application development focus -- I fear that if you take that away, you will go back to what has been happening in the past 10 months. We included these RFP issues to prove that this group could accomplish something, and work. I'm afraid we could not separate out these issues. Tom Mowbray took a straw poll to see who feels this should be handed as a working group. For: a significant number Against: a significant number There was discussion on the scope of this WG and its relation to facilities outside its scope. How can this group write RFP3 if its scope does not cover all the facilities? There was discussion. Sankar Virdhagriswaran said that TF would write it. Maybe we should break up the RFPs differently, into two themes. Tom Mowbray called the question on the amendment. Yes: 17 No: 18 Abstain: 4 Amendment fails. Tom Mowbray called the question on the original, unamended motion. Yes: 19 No: 23 Motion defeated. Tom Mowbray said we can reconsider roadmap questions later this afternoon. Herm Fischer talked about the appdev RFI (draft is document 95-6- 9). Bob Hodges volunteered to be scribe for suggested changes. There was discussion of changes to the document. Jeff Mischkinsky asked if there is a definition of repository here. Herm Fischer said that by asking for a definition, we are helping people to respond. Someone asked what the difference is between repository and information model. Herm Fischer said that's something I hope we see defined in the responses. We want to get people to speak up and say what they think these are. Dan Chang said that we had an RFI before -- it would be nice to incorporate some of the information we received there in this RFI. Herm Fischer said that we have a list of possible functionalities -- but we really can't narrow it down at this point. Dan Chang asked about the scope and focus of the RFI. Herm Fischer said that a year ago I would have said just appdev repository -- now, I think most people would say that they see repositories as being broader scale than that. I think it's more than finding your C program of a given version, getting close even to the running management of a system. Jishnu Mukerji said that over the past year, there have been discussions of repositories in system management and many other areas. There was discussion. Dan Chang said that what's the difference between repository and information storage & retrieval. Herm Fischer said I think that's a good question -- I think we'll be finding that out as we look at responses. To get back the issue at hand, what changes do we need to make to this document? Is there anything that is either too inclusive or too exclusive? We'll be available all day to receive comments on this. Herm Fischer moved to recommend this RFI (95-6-9) to the TC. Taligent seconded. Linda Fisher pointed out a few typos, spelling errors, style issues, and grammatical points. Herm Fischer said that I will look at those. Herm Fischer amended his motion to fix such errors. Informix moved white ballot. No objections. Motion passed by white ballot. Tom Mowbray said that the next order of business is to see a demonstration of Fresco. Mark Linton gave a demonstration of Fresco. Sankar Virdhagriswaran mentioned that one of the advantages of considering all objects to be the same is that you can do scaling, rotation, etc. But it is also clear that most GUI tools don't support that sort of functionality. Eventually we are left with a scenario where we are dealing with a class of objects that cannot take full advantage of the available power. Given that, how useful is all this functionality when dealing with these legacy objects? Is this power valuable to us? Mark Linton said that this is the "can I make progress and still not fall off a cliff" question. This model should work fine if you just want to use the current style. Our model is that over time, we would hope to see that the older applications would become capable of using this power. This is an ongoing issue--I think it is good to move forward. Tom Mowbray said that the editor for RFI3 (Bob Hodges) was wondering if he could add one thing to the document. Bob Hodges said that people have asked if we could include a brief summary of the CORBA Services and Facilities, both for completed and planned services/facilities, with a brief explanation of what each is, to help responders who are not OMG members understand what some of the lower level services that might affect repositories might be. I was wondering if there was any objection to that. There was no objection from the TF. Tom Mowbray said that if there is an OpenDoc demo at this meeting, it will be tomorrow at 10AM. Neil Katin said that there will indeed be a demo. Break for lunch at 11:54. Meeting resumed at 13:07. Tom Mowbray said that we will be looking at evaluations for the Compound Interchange and Presentation Facility. At the last meeting, we realized that the OpenDoc submissions covers both facilities, while the Fresco submission covers only Compound Presentation. Wilfried Kruse presented his Compound Interchange evaluation (slides are document 95-6-24). Neil Katin said that I don't want to respond in detail, but we did produce a characterization of everyones' responses to these evaluations. I could distribute it now, which would probably be more efficient than responding aloud. Tom Mowbray said that sounds like a good idea. (Slides are part of 95-6-24.) Tom Mowbray then presented his evaluation of Compound Interchange. He said that there are two key points to the evaluation. The first point is that CORBA Services are not used in the submission; I identified several places where they could be applied. I know this would take a lot of work to fix; and I'm just pointing it out as an issue. The second point is that overall the architecture was well thought out, and I particularly liked the way that the framework is designed to facilitate development. Neil Katin said that we went through the various evaluations. One wider issue is how to tell when you are done with an object. We use reference counting, which does has its problems. Tim Brinson suggested an alternate method that really is just counting object references, which seems less desirable. Mark Linton said that it's not an either/or choice. The way the spec is done now, you have to do both. I think you have to manage the object references. Richard Soley said that question is how you go about reclaiming storage, which is a separate issue. Mark Linton said that you have to deal with what happens when you don't know if a object reference got somewhere or not, in a distributed environment. Neil Katin agreed -- I don't know if using a counter to do this is the right method, and I think a different question. Mark Linton said that it seems like an OpenDoc application should work even if you forget to unref an object (though it may not perform as well). Someone said that there seem to be terminology problems; however, I'm wondering if you are duplicating functionality that is already there; and second, is this something which is truly needed for Compound Documents. It seems to me this is something that is already covered by duplicate and release. Richard Soley said that CORBA allows you to have any number of references to a given object. There's no model of reference counting. Releasing an object reference does not cause release of the storage occupied by the object. There's no notion of garbage collection in the specification. Perhaps it should have been specified separately. The question remaining is whether or not Compound Documents require this, and the answer seems to be yes. Neil Katin said that there's no way for an ORB to know how many object references are out there for a given object. This is a more general issue. Mark Linton said it's important to distinguish the persistent lifetime of data from the active lifetime of objects. Deactivating is one thing, destroying an object is something else, something a lot more permanent. Neil Katin said that in OpenDoc, we are using reference counting as a mechanism to decide when objects can be destroyed. Mark Linton said you're not using it for persistent things. Neil Katin said this information may be used to make certain objects persistent. OpenDoc uses a "save" model, where things in memory are eventually made persistent. When you do reference counting, you are just deleting in memory copies. Eventually, however, the memory representation will be externalized. Mark Linton said this is probably worth having a subset of interested people talk to ORB or OS people. It seems to me that servers have to have some reclamation strategy to handle garbage collection. Neil Katin said that we are leaping from interfaces to implementations. Mark Linton said that explaining what you should do in terms of failure often affect the interfaces. Neil Katin said that maybe we have to examine the failure cases for our current set of interfaces. Someone asked if I need to ref an interface, do I also do duplicate at the same time? Neil Katin said yes -- things that bump reference counts don't necessarily need new objects. Some of them pass you object references directly. There was discussion. Neil Katin asked how important IDL naming style is to people. We're thinking about using modules but not changing our naming style. Someone said that modules are definitely more important -- on the other hand, this is the first facility, which will likely affect the standard for future facilities. Richard Soley said that you have no choice but to use modules. In terms of style, if this is adopted, I think it will look very different from everything else that has been adopted, and that will be bad. Borek Vokach-Brodsky said that this is an issue that will come up again and again. Richard Soley said that it's not something that is hard to fix. Neil Katin said that it's hard to change customer code. Richard Soley said that it's easy to have two names for an interface. It would be good to have consistency. Someone said it's easier to fix now than later. There was discussion. Mark Linton said that I think the feedback is that it would be really good to have a consistent naming style. If you think it's too high a cost, I think you will have to make a case to the TF. I do think we need to determine the prefix for CF modules. Richard Soley said that in previous adoptions, naming conventions were by the submitters and the editors of the document -- it doesn't matter what the specifics of the conventions are, but they needs to be agreed upon. Neil Katin said maybe we can talk about it this afternoon. Jeff Mischkinsky said that I think it's important for OMG to present a consistent view in its interfaces -- I think this is very important. Neil Katin said that we don't use attributes -- we think they are architecturally broken -- live with it. Richard Soley asked if you are using attributes where there are no user-defined exceptions. Neil Katin said that we at current don't have user- defined exceptions period. Jeff Mischkinsky said that I don't think it's a big deal. Neil Katin said that OSA and AppleEvents will be deleted in the next revision of the submission. Dan Chang asked about the storage protocol. The submission talks about stream-only storage--will you be addressing that? Neil Katin said that is a case where our implementation shows through. We'll deal with that in the next revision. Sankar Virdhagriswaran presented his evaluation of OpenDoc. Tom Mowbray said that there were a number of other people who signed up to do evaluations who did not. There's one more evaluation, by Guus Ramackers of Oracle, who is not present at this meeting. Dan Chang asked about the storage system. Neil Katin said that out of the box there are two storage systems: OLE and Bento. There's another example of how the implementation that is being shipped show through to the interfaces, which is a mistake. Tom Mowbray asked if OLE integration is part of the CI Labs release. Neil Katin said yes. Tom Mowbray said that we've gone through the written evaluations -- are there any verbal comments? Dan Chang said that OpenDoc also related to Transactions, Concurrency Control, and Properties Service. Sankar Virdhagriswaran said that I raised a number of problems -- I would like to understand how they are going to be addressed. Neil Katin said that we didn't have time to write down detailed answers, or to discuss them here. The things we would most like to talk about here are issues that require group discussion. Tom Mowbray said that we are not tightly bound timewise this afternoon. Richard Soley said that there are clearly questions that Neil Katin cannot answer here on the fly. Borek Vokach- Brodsky said that some of the questions raised, for example modules, simply will be addressed in the revised submission. Tom Mowbray said we will move onto compound presentation facility. Tim Brinson presented his evaluation (slides are document 95-6- 25). There was discussion on rendering issues. Sankar Virdhagriswaran said that there are problems with interoperation between parallel services. There's an OpenDoc naming service, event service, etc. Neil Katin said that we've already said we want to add connections to our submissions. Talking about rendering, once the CF TF decides how rendering works, you can plug it into our facility. In the meantime, we had to come up with something -- we have it so that you can take whatever rendering mechanism you want and plug it in. Tim Brinson said that Fresco has rendering now, and that provides a level of distribution. If you want to use OpenDoc for standard rendering, you have to go outside the facility. Tom Mowbray said that there were other people who were supposed to produce evaluations. Would those reviewers like to present any verbal evaluations? Dan Frantz said that I want to make sure that gratuitous references to SOM and Macintoshes will be left out. The IBM submission says in a couple places that OpenDoc is sitting on top of SOM and is therefore CORBA compatible. I want to make sure that you're CORBA compatible and not SOM compatible. Neil Katin said that there is an OpenDoc implementation and an OpenDoc specification. The implementation sits on top of SOM. We've tried to avoid SOM-specific features, but there are some functions which we had to have. We believe that it wouldn't be hard to port OpenDoc to another ORB, but until we've done it there's no way to know for sure. Richard Soley said that I believe that SOM has a flag to produce only CORBA-compliant code. Neil Katin said that we've tried running our IDL through other party's compilers. If there are problems found, we are committed to changing the implementation. Geoff Lewis said that from the spec, you could produce OpenDoc not using SOM. Neil Katin said that I would hope you can implement the spec on any ORB. I was talking before about our body of code. Geoff Lewis said that if that was done, what does it do the spec's portability. Neil Katin said that it's hard to answer in the general case. What is the API difference between those ORBs? Right now you can't write an object that works on different ORBs for the parts that have not been standardized. Geoff Lewis asked how much OpenDoc relies on the class mechanism of SOM. Neil Katin said zero -- it does not rely on it at all. Our design goal was to use none of the proprietary extensions of SOM. The things that are ORB specific are the things that are not OMG standardized yet. Someone asked if you have distributed it across machines. Neil Katin said that we've done a little prototyping across machines. We've done it across processes. I think we've answered the programming question -- what we haven't answered is the performance question. Tom Mowbray said that the OpenDoc team submitted material covering the End-User requirements. Neil Katin said that the End- User SIG people will probably be more interested than the people here. They've already met for this week; I will get this to them separately on the mailing list. Break at 14:36. Meeting resumed at 15:05. Tom Mowbray said that the next thing on the agenda is how to follow-up on the RFP1 evaluations. RFP1 Evaluations - Open Issues: Response to questions from evaluations Response to Compound Presentation questions formulated in Cambridge Response to End-User scenario Wilfried Kruse said that discussion here might be better than a written response for some of these issues. Martin Anderson said that I don't think the EU scenario is a distribution issue, so much as things the submissions may not be attempting to go after. Tom Mowbray said that Richard Soley suggested asking the submitters to write responses and present them at the next meeting. Mark Linton said that I thought questions 1 and 2 were for submitters, and question 4 was for evaluators. [See Cambridge minutes for a list of these questions.] Sankar Virdhagriswaran said no, question 4 is for submitters. Tim Brinson said that an additional part I was reviewing was reviewing technical details of each submission; however, I have not yet completed that. I'd like to get them to the submitters at some point. There was no objection from the group. Neil Katin asked for a deadline. Tom Mowbray said it's not our intent to re- open evaluation per se. Neil Katin asked if we should present the EU scenario to the TF or the EU SIG? Tom Mowbray said it could be presented here and we could invite the EU SIG. There was discussion. Mark Linton said it's a very easy thing to answer, and I'm not sure it's worth spending much time on here. I think it would be better to focus this answer on the EU SIG. Tom Mowbray said that I will see if I can coordinate a joint session with the EU SIG. Neil Katin said that it's very hard to respond to these things because the questions are malformed. They're not properties of the standard - - they're properties of a particular implementation of the standard. Paul Murray said that the End-Users understand the problem, and what you just said. Personally, I would not want to spend too much time on it. Tim Brinson said that there was a lot of discussion about this in EU SIG meetings. This is the first RFP on which they had a requirements appendix. When I first read the appendix, it didn't really say you had to respond directly -- I guess they were trying to get an explicit response. Somehow they developed this scenario -- I was in EU SIG meetings and didn't see it, so I don't know where it came from. Paul Murray said that it's difficult to figure out how to get the End-Users to form the questions well. Neil Katin said that let's have a joint session or have selected people attend the EU SIG and try to get well-formed questions. There was discussion. Mark Linton said I'm lost as to where we are with object services. I think we should have some formal interaction with the people who wrote those services. Tom Mowbray said that there are many of those people in the room. Mark Linton said we can do it informally--I just want to make sure we sit down and talk. Neil Katin said I don't understand why we need to do that. Mark Linton said that meeting with people who are more knowledgeable would be fine. Neil Katin said that I think that would be best to take care of outside the OMG process. Mark Linton said that I don't care how it's done -- I just want to make sure it's not dropped. There was discussion. Rudolf Riess asked about timeframe. Tom Mowbray said that we just finished discussing questions that the submitters will answer in Ottawa. Revised submissions are due 25 October; we can discuss them at the November meeting and vote a recommendation. Tom Mowbray said the next order of business is to talk about the naming prefixes. Proposals: CF CCF [later removed] Cf Ccf Tim Brinson moved to use 'Cf'. Crystaliz seconded. Neil Katin asked if we could do a straw poll first. Straw poll: Cf: 8 Ccf: 0 CF: 11 Neil Katin proposed amendment to capitalize the 'f'. Sankar Virdhagriswaran seconded. Someone suggested voting between the two instead of the current motion. Tim Brinson agreed to that change. Tom Mowbray called the question. Cf: 11 CF: 5 'Cf' passed as the Common Facilities prefix recommendation from this breakout group. Tom Mowbray said that we are now talking about a general style setup for module names. There was discussion. Neil Katin asked for those in favor of using the name that appears in the CF RFP. Yes: lots No: few Neil Katin said now concretely, is it a spelled out name or an abbreviation. Mark Linton said that it sounds like you are precluding the thought of multiple modules in a single facility. There was discussion. Someone said that I think we can leave it to the submitting teams. Tom Mowbray said that we could look at the existing submission and generalize from there. There was discussion. Someone suggested adding this to the submissions. Carol Burt moved to have the submitters propose a naming scheme at the September meeting. Sankar Virdhagriswaran seconded. Borek Vokach-Brodsky said I think the timing is way off. If the CF rejects our proposal, we won't be able to change our submissions before the revised deadline. Wilfried Kruse suggested keeping it as part of the actual revised submission. Rudolf Riess proposed an amendment that we allow the submitters to choose whatever name they want for the modules within the limitations: only generic names and names that reference what the model does (and no trademarks); and that these be included in the revised submissions. Neil Katin seconded. Carol Burt accepted the amendment. Tim Brinson moved white ballot. No objection. Motion passed. Break at 15:58. Meeting resumed at 16:09. Tom Mowbray said that we need to talk about the roadmap and the next RFP. Start RFP Items Planned adoption Oct94 RFP1 Compound document Oct95-Feb96 Mar95 RFP2 Internationalization & Time Operations Mar96 Sept95 RFP3 Data interchange & Automation Sept96 Sept95 RFP4 Scripting agent facility, Rule Management Sept96 Feb96 RFP5 Information Exchange Feb 96 Tom Mowbray said that is has come to my attention that the readiness of these technologies varies, and that the division we did on the roadmap does not make sense today, though it might have a few months ago. In particular, I know companies are ready to begin work on data interchange and information exchange; however, it doesn't seem that companies are ready for automation. Tom Mowbray presented a listing of RFP3 candidates: Data Interchange Automation Scripting Agents Rule Management Information Interchange Sankar Virdhagriswaran asked who is chomping at the bit to respond. Tom Mowbray said that there are multiple companies in the Business Object SIG, as are Dan Chang and myself. Dan Chang proposed Data Interchange and Information Exchange for RFP3. SSA seconded. (Systems Software Associates). Sankar Virdhagriswaran suggested that Data Interchange be in the RFP and Information Exchange not be in the RFP. Sankar Virdhagriswaran proposed a friendly amendment to have the RFP3 editors work on Data Interchange, and let them decide whether or not Information Exchange should be added. Dan Chang accepted the friendly amendment. Claude Baudoin asked if this would be a plenary activity rather than a splinter group. Tom Mowbray said yes. Craig Thompson asked what Information Interchange is, in a nutshell. Sankar Virdhagriswaran said I want to know what data interchange is. Craig Thompson said that I think data interchange covers transfer of data not in IDL formats--getting data in and out of an OMG environment, and from other environments. Tom Mowbray said that there is a facility description of both of these in the architecture now. The primary difference is that information exchange deals with metadata and semantic mediation issues. The assumption is that both facilities are specified in IDL, and each can handle both IDL defined internalized types, as well as externalized data, for example, conveyed as sequences of octet or char. There was discussion. Tim Brinson said that the original roadmap had two RFPs coming out at the same time. Is that still the case? Tom Mowbray said that's a separate issue--we still have that option. Dave Zenie asked if by data interchange we mean taking of data and file formats and translating them. Tom Mowbray said yes, although I don't think it would mandate translation of information to any specific formats. Tom Mowbray called the question. No objections. Tim Brinson moved white ballot. No objections. Motion passed. Tom Mowbray asked if there are any other orders of business. Tim Brinson said that we still don't have interoperability between systems, so I think Rendering needs to be addressed as soon as possible. Paul Murray said that there was an attempt this morning to start a working group that would be based on coming up with more focused requirements that would perhaps speak to the specification. That working group would have been helpful in this discussion. That was an excellent opportunity, but it was voted down. I just wanted to point that out. Sankar Virdhagriswaran said that mission was too broad. If it had been stated exactly the way you said it, it would have been different. But the mission made it an ongoing working group. There was discussion. Paul Murray said that there were people who were willing to do the work for free...no skin off anyone else's back. Craig Thompson said that anyone is more than free to help us with the work--there's no need for a special group to have end-users help with the TF. Herm Fischer said we are looking at an RFI for Distributed Development Tooling and for Component Assembly. Tom Mowbray put up a list of candidate RFP4 facilities, these include all the remaining facilities from the CF Roadmap and Rendering as suggested by Tim Brinson: Rendering Scripting Automation Agents Rule Management Tom Mowbray asked if there were any other candidate facilities that we should consider for RFP4. Dave Zenie said that we may be submitting an RFC on workflow. Herm Fischer said that the Workflow Coalition is looking at ODBC -- does that impact any of these facilities? Sankar Virdhagriswaran said that it's covered by Query Service. There was discussion. Rudolf Riess asked if there are submitters eager to answer RFPs for any of these facilities. SSA moved to have RFP4 include Rule Management and Agents. Dave Zenie seconded. Sankar Virdhagriswaran proposed an amendment of adding Scripting. SSA accepted. Craig Thompson said that if the timing is exactly the same, won't there be some unnecessary overhead by packaging them separately? Tom Mowbray said that there would be more coordination overhead by combining them. Tom Mowbray said it would take a 3/4 vote to reverse our decision on RFP3. Tim Brinson asked if the review cycle can be separate or together? Tom Mowbray said it doesn't matter--we can have the review cycles go however we want. Tom Mowbray asked who would be interested in editing this RFP. Sankar Virdhagriswaran and Craig Thompson volunteered. Tim Brinson asked if this precludes yet another RFP. I would like to push Rendering. I'd like to get a sense of the room whether it would be better in here or as a separate RFP. Sankar Virdhagriswaran said that Rendering got pushed last because we didn't know what it would look like. Tim Brinson said that we've seen the Fresco submission since then. Tom Mowbray suggested that another facility would add even more to the workload. Tim Brinson said he would write the RFP. Tom Mowbray said that the review cycle was what I was concerned with. Tim Brinson asked if we are going to start doing RFPs for pieces of facilities. There was discussion. Tom Mowbray asked for a straw poll as to whether Rendering should be added to this list or as a separate RFP. Added: 0 Separate: lots Rudolf Riess proposed an amendment to include Automation. Sankar Virdhagriswaran asked what is meant by automation. SSA rejected. Rudolf Riess withdrew. Tom Mowbray called the question. Yes: 12 No: 1 Abstain: 1 Motion passes. Tom Mowbray asked if there was any other business for today. Carol Burt said that I think it would be a mistake to wait too long for a Rendering facility. Tom Mowbray said that the ORB/OS TFs decided that Messaging should probably be placed in the OSTF. The chairs of the two TFs will be talking to us tomorrow about that. Tom Mowbray went over the agenda for tomorrow. Break at 16:44. Thursday, June 29, 1995 The meeting was convened at 10AM There was a brief discussion of the messaging object service by Geoff Lewis and Andrew Watson (OSTF AND ORB TF Chairs). Following this discussion, there was a demonstration of OpenDoc by Raj Kanodia, Apple. The demonstration comprised an exposition of the OpenDoc technology, in support of the OpenDoc technology submission to CF RFP1. The meeting was adjourned by 11:15AM.