See also: IRC log
<scribe> scribe: Noah Mendelsohn
<trackbot-ng> ACTION-106 -- Norman Walsh to make a pass over the WebArch 2.0 doc't which adds a paragraph, and connects up to issues list -- due 2008-05-08 -- OPEN
SW: We've talked about this before. Norm reports no progress on ACTION-30.
NW: That's right, I'm sorry I didn't get to it.
SW: I'm not sure how we get started.
<trackbot-ng> ACTION-106 -- Norman Walsh to make a pass over the WebArch 2.0 doc't which adds a paragraph, and connects up to issues list -- due 2008-06-05 -- OPEN
NM: Calling the discussion "Volume 2" prejudices the options a bit. We may want to do a lot of our work in a new edition of the current volume. The ultimate result should be document(s) that readers perceive as well organized.
JR: We don't want to do technology for technology's sake, or writing for writing's sake. Would doing a use cases and/or requirements analysis be one way to focus the goals?
SW: We have a skeletal outline.
(Norm goes to look for it.)
SW: We've had substantive discussion of Semantic Web and of clients with richer application behavior. There's also the question of doing maintenance work on what we've already done.
JR: Maybe we should be goal based. Is the goal to influence certain groups to do certain things?
TBL: Describing the Web as a system, from the top level and as an integrated story, is important. The reason the topic areas seem disjoint is that having started sort of top down on the table of contents for Volume 1, we found that some areas were contentious and some not. Still, it's important that what we're shooting for is to tell one consistent story.
Norm finds the list of topics at: http://lists.w3.org/Archives/Member/tag/2008Apr/0058.html
JR: The purpose of the finding is to convince certain people to do certain things.
NM: I think the audience is broader than one might think. For example, I often use the Architecture Document to educate people who are not coding applications, but who are making decisions like whether their mobile phone applications should integrate with the Web. From that they learn that they'll have to tell their programmers to identify things with URIs, etc.
JR: The question is how do you decide which project to work on.
TBL: Maybe the top level document should say the obvious things without very much depth. Consider naming. It's important, but not separate from other aspects of the Web. Viewed from one perspective, the Semantic Web is just one format or set of formats. There are other questions about how AJAX works with the Web.
JR: Maybe use cases, as Noah has been talking about.
NM: I think we sometimes have to publish the things where we've been able to make some progress. Sometimes we set a few priorities, but then find that we can only generate good insights into some areas we'd hoped to hit. Which ones will progress is not entirely predictable. So, we should at least consider sharing the things where we've made good progress.
NW: I think that the business of working in a document, producing drafts, actually focusses our work. Also, looking at mining the information in the findings can be useful.
TBL: The TAG was formed in part because working groups used to be told "you can't do that, it's not how the Web works", but there was no common point of reference for how the Web really does work. Now that we have the mandate, we work in two modes: sometimes we are responsive to an external question or an issue that arises, but sometimes we are proactive in setting priorities. When we produced Web Architecture Volume 1, many people told us it was useful. So, we weren't just satisfying ourselves. This time, I would like to see more formal methods, though.
JR: I still think we could do a more careful job of capturing those goals and dynamics.
SW: I think there are three bullet points in the TAG charter that pretty much capture what Tim said.
<Zakim> timbl, you wanted to mention origin of the TAG in WG questions
<Stuart> Mission statement
<Stuart> The mission of the TAG is stewardship of the Web architecture. There are three aspects to this mission:
<Stuart> to document and build consensus around principles of Web architecture and to interpret and clarify these principles when necessary;
<Stuart> to resolve issues involving general Web architecture brought to the TAG;
<Stuart> to help coordinate cross-technology architecture developments inside and outside W3C.
<Stuart> From http://www.w3.org/2004/10/27-tag-charter.html
DO: I like a lot of what we've talked about doing either in Vol 2. or updates to Vol. 1. Doing versioning and/or self-describing Web and/or distributed extensibility would be really great, but.... I still feel as I've said before, that there's a huge buzz around AJAX, social networks, and the "latest cool things", and it's not clear to me that we're doing a good job of helping them. Not sure what to suggest, but it feels like it could be a good goal to really align better with where a lot of investment and energy is going.
<Zakim> DanC_lap, you wanted to noodle on events, e.g. Future of Web Apps
DC: I haven't heard anyone talk about usefulness of Web Arch. vol 1 in awhile. So, I'm interested in focussing on events like the Future of Web Apps. There's likely to be discussion of some things like the future of Twitter is. Would anyone like to talk about interesting conferences and how we could both have impact and be influenced by their needs?
DO: Yes, that would be great.
DC: We had some presence at XTech, and that seemed good.
DO: Supernova's coming up in San Francisco in June.
NW: Supernova 2008 (http://www.supernova2008.com/)
<DanC_lap> (hm... http://www.supernova2008.com/ )
NM: Should we go through the details on these things here, or agree to do in email?
DC: I prefer here.
NM: I'm still enthusiastic about the idea of doing a piece of a TAG Web site that would be a really cool, useful tutorial and introduction to why Web architecture matters, and how to apply it.
SW: Books, such as Web Architecture in a Nutshell or something of that ilk would be useful.
JR: Not sure we can pick up much more work.
DC: The nice thing about blog articles is that we don't have to agree before publishing.
NM: Yes, we've decided to use our blog that way, and it seems to be proving useful.
<DanC_lap> that wasn't my point:
SW: I need guidance as to how we make progress.
DO: The issue has come up about walled gardens like Facebook in which things are not properly identified with URIs.
<DanC_lap> (jar mentioned reviews in blog items... I tend to microblog in bookmark services... I have ~180 in http://del.icio.us/connolly/architecture )
NM: Engaging folks like Facebook would be great, if not necessarily easy on a subject like this, but I'm not sure how it informs what we'd write in a new edition of WebArch.
TBL: One of the social network issues is: how does my social networking site get a list of friends from your site? One good answer of course is, it should be retrieved as a document probably using a technology like RDF. Then you find that you need what can be rather complex 3rd party authentication systems like oauth. Interesting that so much of this is being driven by the "friends list" use case. At the moment, it's not being done RESTfully. Now, all of this is being done in service to building applications that integrate data from two or more social networking sites. You can imagine doing this in the browser if you like.
DO: Oauth is pretty cool in how it uses URIs.
TBL: It does a lot of redirects. We looked at OpenID and the number of round trips is large. Oauth seems to do even more. Popping up: it would be really interesting to ask "how would you do something like this better using Web Arch"?
DO: What would that look like? What would we on the TAG do? Look at OpenID and oauth from a Web Arch perspective?
TBL: Yes those, but also the larger problem of portability of things like friends lists.
NM: I'm not sure what to do about it, but if the sorts of applications one sees with Flash and Silverlight become more and more ubiquitous, we might ask what if anything Web Arch Ed. 2 should do to help you know how to build apps with that level of function (not necessarily using those proprietary technologies). Otherwise, there's some risk that by the time we publish, we'll be talking about applications that look a bit old fashioned.
HT: Often what we're doing can be best viewed as industrial archeology. We often do best when we look back. So, Noah, I'm not sure those rich technologies belong in our document.
NM: I agree actually. I tried to signal that I too think we often do best when we look back, but it runs the risk that by the time we say anything we don't influence the people who can still make choices about new things.
DO: Someone has to figure out how to "put pen to paper". Once it gets done, I would be interested in putting a subset of the compatibility strategies into Web Arch vol 2. For example, Web Arch v.1 says "use version ids", and we've decided the story needs to be more subtle than that.
SW: (Stuart shows a picture of a useful bridge over a river that's made in cast iron, using a design meant for wood). This is a good example of how architecture does not always come first.
... I got this from a talk by Peter Williams, who used to work here at HP.
SW: Perhaps we need to start by doing a thorough review of Web Arch. vol 1?
JR: Aren't the revisions a separate bit from the new stuff?
SW: Revising is expensive, so doing it for one piece isn't a good bet.
<Zakim> Stuart, you wanted to pick up on industrial archeaology
NM: I think we need to focus not so much on separating revisions from new stuff, but on creating a document or documents that will serve the community for 3-5 years. We'll find out what's revised, what's new, and how it's best organized as we go, I think.
<trackbot-ng> ACTION-106 -- Norman Walsh to make a pass over the WebArch 2.0 doc't which adds a paragraph, and connects up to issues list -- due 2008-06-05 -- OPEN
NW: I have an action to add details to our working list of possible priorities and to link them to our issues list. I suggest that the next time we should discuss Web Arch future is after I complete that action.
SW: I suggest we do as Norm suggests.
JR: I wonder if we need to collect things other than what are in http://lists.w3.org/Archives/Member/tag/2008Apr/0058.html
SW: Yes, but it's a start. We'll change as necessary. I think that some of figuring out what needs to be done comes of reviewing what we've already done. I think we've heard that reviewing the written works like findings and AWWW vol. 1 is something that individual members will do as they see fit.
JR: Where do we do this?
SW: At least in www-tag, maybe sometimes in the TAG blog.
<DanC_lap> close action-114
<trackbot-ng> ACTION-114 Henry to find the counter example that made it necesseary to make a terniary relationship closed
<DanC_lap> close action-115
<trackbot-ng> ACTION-115 Henry to improve the presentation of the way the ontology reconstructs RDDL 'purpose', and to attempt to address skw's concern about the subject of the so-called purpose relation closed
HT: Maybe we need to talk about how to get HTML and XML to converge? We have Sam Ruby's distributed extensibility proposal ( http://www.intertwingly.net/blog/2007/08/02/HTML5-and-Distributed-Extensibility ), the IE8 proposed function, etc.
<DanC_lap> ^ sam ruby on distributed extensibility
HT: We've also talked about implicit namespace bindings. These could be implicit in the media-type, perhaps written into the specifications e.g. if svg; aria: prefixes were "written into" the HTML specification. So, these are things that led me to suggest putting this back on the agenda.
<DanC_lap> Maciej Stachowiak
TBL: Someone (Macieg Stachowiak?) has suggested having the occurrence of an element implicitly bind a namespace. Another option would be to declare in a specification that you can find by following your nose that, e.g., in HTML, the occurrence of a certain element such as SVG would trigger the binding.
... Alternative would be to have "svg:" prebound in HTML.
... The idea is to get a smooth slope, where millions of people can do the easy thing easily, but it scales architecturally to the general case with the full URI.
<timbl> Got a pointer?
<timbl> I feel it is important that the HTML5 spec be split into smaller chunks.
TBL: I feel it is important that the HTML5 spec be split into smaller chunks.
DO: I think Roy Fielding has made similar comments.
SW: We received an invitation from Ian Hickson to review the HTML 5 spec: http://lists.w3.org/Archives/Public/www-tag/2008May/0041.html
HT: What's the right way for us to say that it's going to be very challenging to review something that large? If we could find a way to agree to focus on more specific pieces, that might be helpful.
TBL: I think Roy's feedback was fundamentally "where's the core language definition?" I think that would probably be roughly mine too.
DC: I want to send comments primarily where it's likely to produce useful action.
<DanC_lap> found Fielding's remarks that DO alluded to: " This draft has almost nothing to do with HTML. It is a treatise on browser behavior. That is a fine standard to have, but deserves a different title so that the folks who just want to implement HTML can do so without any of this operational/DOM nonsense." -- http://www.w3.org/2002/09/wbs/40318/wd11spec/results
DC: I'm trying to figure out how many MIME types there should be for HMTL. Some people believe there should be application/xhtml+xml in addition to text/html. There's a point of view I might share that xhtml+xml isn't going to take off in practice, but some people feel very strongly we need to make xhtml+xml work.
Poll on how many HTML media-types we should have in the long run:
DO: strong 1
NM: bits of opinions, but not coherent or informed enough, so I suppose I pass
<DanC_lap> (I realize this poll is more about tagSoupIntegration than distribute extensibility)
HT: Now or in the long run?
DC: Long run.
HT: 1, but only the very long run.
NM: Not sure I have a well informed opinion (noodles on saying just text/html)
HT: Um, not sure. As long as XHTML 1.n has any traction at all, I want an XML media type for it. Because I want what follows from that.
<Norm> In an informal poll, mostly what Henry said, 1 if it's application/xml+html. If I can't have *that* one, then two. I guess.
HT: Until we have some single thing that is really "both", with good statutory grounding.
NW: Mostly what Henry said. Ultimately 1, if I could make it the XML one.
DC: If application/xhtml+xml, it will have to do all the non-clean XML stuff, so text/html (1)
JR: If the distinction gives you information that's useful early that's good, but I'm not sure I care as long as whatever we get supports "follow your nose".
<DanC_lap> DC: FYI, Sam Ruby argues for 2. http://www.intertwingly.net/blog/2006/12/18/application-xhtml
TLB: Convergence is really important between HTML and XML, because there's only one HTML. I'm not sure that having no "+xml" in the mime type explicitly is something one can work around as a special case, I.e. interpreting text/html as XML when necessary. I don't resent +xml, but I think text/html should migrate smoothly to XML over time. BTW: text/* should perhaps migrate to UTF-8.
JR: What's the nature of the distinction? You're not going to treat the text differently. You can get the information from 3 places: from the media-type, from the start of file. If you can heuristically do a parser that "does the right thing", that might work.
NM: If the media type spec. delegates the "is it XML?" switching to what's at the head of the file, then OK, otherwise not. You can't do it, e.g. for text/plain.
JR: I wonder if 2 is the wrong number.
<Zakim> ht, you wanted to mention the difference between RDF and SVG
<Zakim> noah, you wanted to talk about mixins in media types
<trackbot-ng> ISSUE-9 -- Why does the Web use mime types and not URIs? -- CLOSED
NM: I tend to feel that one architectural issue is the lack of richness in the structure and semantics of media type names. If you had mixin syntax and semantics, you could come closer to saying "oh, by the way, this text/html or this text/plain happens to be well formed XML, and you have permission to interpret it as XML."
HT: RDFa is an example of a small vocabularly of which you can say "it's not obviously wrong to say that you should just negotiate with the HTML WG to make them part of the HTML base language". RDFa has a handful of elements, ARIA, more, SVG has a large number of elements and MATHML more. All are enumerated in the specs.
... If I were approaching this a priori and adding RDFa to my <spans>, I would like to write <span my:foo="xxx"> where my:foo is an RDF relation. What's interesting is that my:foo's are open-ended.
... It seems to me that the tail has been wagging the dog in designing RDFa. In a sense, the reason we're not seeing a strong example of the need for distributed extensibility is that the lack of namespace-based extensibility in HTML bounded the discussion to require at most a few new attributes,
DC: Your view of the aesthetics may be in the minority. DTD-based validators are part of what drives this. Also, people who work with markup languages believe that tags are "holy".
HT: Attributes sometimes feel different.
NM: One thing you get from always triggering a capability (e.g. RDFa) with well known attributes is that a client is more likely to know about a well-known "property" attribute than a domain specific foo:xxx. It it will at least know that what you were trying to do was RDFa.
HT: RDFa wants support in browsers.
NM: I think that whether there are various RDFa add ins to browsers or built in support isn't the point. It gets built in iff it so happens that the same function is needed by a large class of users. If lots of users need different variations on the UI or functions for handling RDFa, then selectable addins are right.
SW: The relationship to HTML 4.01 XHTML 1.1 DOM2 HTML section in the HTML 5 draft positions HTML 5 as successor to XHTML. Yet, it seems not carry forward capabilities like modularization that depend on namespaces.
HT: The XHTML modularization spec uses a trick that you (Dan) and I discovered independently to allow flexibility of prefixes with DTDs.
DC: Adequate is in the eye of the beholder. The workgroups seem to disagree. Some people believe that all that happens is that attributes were added, and the availability of a fancy schema to describe them is secondary.
<Zakim> jar, you wanted to say metadata ( link header) can do what media types can't
JR: Metadata in link headers might take some pressure off the urge to enrich mime types.
<DanC_lap> (hmm... I thought Adjourn meant the end of the whole meeting, but nope... "To adjourn means to suspend until a later stated time." -- http://en.wikipedia.org/wiki/Adjournment )
=========== RESUMING AFTER LUNCH ============
<Askok> scribe: Ashok Malhotra
SW: There was some discussion of versioning -- should we allocate time for it?
HT: We should allocate time for JAR to present.
Passing baton to Dave and Henry...
HT: We sent msg to ARIA folks. Dave drafted a doc for the tag blog and asked me to look at it. I spent a fair bit of time on the point that XRIs allowed chacraters that are not allowed in URIs. This proved to be a mistake. So, HT has not reviewed Dave blog entry yet.
DO: I can wait till tomorrow.
HT: I'm going to update the finding using the doc I prepared for Vancouver.
<trackbot-ng> ACTION-33 -- Henry S. Thompson to revise URNsAndRegistries-50 finding in response to F2F discussion -- due 2008-03-27 -- OPEN
NM: Introduces the subject. Machines and technologies continue to get more powerful, and UI's sooner or later exploit that. Where once images were an enhancement to the Web, they are now commonplace, but video is new. The machines are now capable of doing video, animation etc., and we are seeing the emergence of Flash, Silverlight, etc. These provide rich animation, multimedia, etc. Note that some of the applications are specifically media-focussed: MS persuaded NBC to broadcast the upcoming Olympics to the Web using Silverlight. The motivation, at least in the case of Microsoft, is claimed in part to be as a lever for getting into Web advertising (cf. Ray Ozzie keynote at Mix '08 conference.) Overall, these technologies are pushing the Web towards new types of apps and interfaces. Right now we don't have a standards-based approach with the capabilities of these platforms. This raises significant issues for the Web. More and more content being creating in these proprietary formats. They deliver over HTTP but tend to violate the Rule of Least Power (they tend to deliver as byte codes, not declaratively, though in the case of Silverlight if you look hard enough the proprietary but declarative XAML can usually be found burried in the executable of a Silverlight application). Often with these applications, you cannot copy/paste to clipboard, cannot do view source, etc. So, in those respects, a step backwards for the Web.
<DanC_lap> (let the record show that the web-with-images came BEFORE the text-only web; it's just that the text-only web got more widely deployed than the NeXT client in early days)
<Zakim> DanC_lap, you wanted to note that the new flash client has p2p support
DC: Adaptive video streaming over HTTP. Chop up video into small bits. Move Networks does this. Strategic relationship with MS.
DC: Network protocol is going on among proprietary players.
SW: ISP are being stressed by video content and want more money.
<Zakim> jar, you wanted to be random (threats, technical fixes, systematic problems)
JAR: We are acting on behalf of the public. What might we do or say?
<DanC_lap> aha... found the flash/p2p item...
JAR: The technical part to fix these problems is not hard.
<Zakim> noah, you wanted to respond to Jonathan
<noah> Quoting from the TAG Charter: "The mission of the TAG is stewardship of the Web architecture. There are three aspects to this mission: 1. to document and build consensus around principles of Web architecture and to interpret and clarify these principles when necessary; 2. to resolve issues involving general Web architecture brought to the TAG; 3. to help coordinate cross-technology architecture developments inside and outside W3C." I think that working with these outside organizations to make their technologies better citizens on the Web is squarely within the scope of #3.
NM: We could do a TAG finding that these technologies lack some features. The W3C could line up and discuss how to make these technologies better citizens on the web.
DO: I think it's worth doing something.
AM: Should we start standards efforts in these directions?
NM: If we can involve people who are in a position to deliver to enough of the community to get critical mass. I don't think I see that yet with some of the existing bits and pieces like SVG. I especially don't see it near term when you consider the sophistication of the runtime and tooling technologies that would be needed to create applications at all comparable to what the proprietary offerings are doing. Of course, I'd love to see a standards-based answer: just that we have to be realistic about whether and how it gets done.
DC: Disagrees -- standards are low risk. They work best when market has figured out where it is going
... The big deal I see is some combination of authoring tools and teaching people to author.
NM: (Note from the editor of the minutes: I'm fairly sure that Tim asked Noah whether widgets were available for Flash and/or Silverlight): Yes, Flash has a rich set of widgets as far as I know. In the case of Silverlight, early releases did not have them, but later ones do, and users can subclass those to create their own.
<Zakim> ht, you wanted to point to old blog entry
??: The basic Web story is dependent on the resource you get whan you do a GET.
??: This is attentuated by new technologies ... whole story is being compromised
<timbl> (Noah, By the way, http://www.w3.org/DesignIssues/diagrams/arch/follow.svg )
??: Threatens the foundations of the Web model
<Zakim> DanC_lap, you wanted to note the least power issue in DVD-next-gen standards: http://zbowling.com/blog/2007/10/08/including-java-in-standards/ <-
<DanC_lap> "Blueray requires an implementation of Java on all Blueray players to run the interactive menus on those systems ... HD-DVD’s menu system is stored in a standardized document format"
DC: Least power shows up in DVD standardization.
<Zakim> Stuart, you wanted to mention WAF Access Control for Web Resources.
<DanC_lap> HT and TBL note that HD-DVD has folded
SW: WAF talks abt access control of resources by resources.
HT: Ties in with idea that Browsers are not the only user agent. Bits that are on the page have less and less to do with what the user sees. GOOGLE is putting huge effort into image recognition. Page rank algorithm has changed and no longer counts linking nearly as heavily as in the past. Olden days most pages were hand authored. Today most pages are synthesised more or less automatically
<ht> I believe that google's ranking of pages today makes relatively little use of the original incoming-links-based 'page rank' number
TimBL: Hypertext analogy does not work.
<Stuart> What I said was about lack of webarch vocab and coverage of user agent threads of behaviours. WAF document speaks of controlling access to resources by resources, but that the thing actually performing access is a thread of execution in a UA rather than the first party resource from which the page was loaded.
<ht> HST remembers to point people to Sean McGrath's XTech closing keynote. . .
TimBL: There is a lot of procedural code to do what they want to do. We could encourage them to be more declarative
<ht> Lots of food for thought here: http://assets.expectnation.com/15/event/3/Orangutans,%20Oxen%20and%20Ogham%20stones_%20Mulling%20the%20movable%20Web%20Presentation.pdf
NM: URIs are not always as separable as you think.
SW: Can talk about concrete next steps.
<ht> Particularly page 63
DO: We could talk about why walled gardens are bad.
<DanC_lap> (I think "walled gardens" like facebook are an important part of the marketplace)
DO: We could do usecases re. Flash and Silverlight.
NW: That's a great place to start.
<timbl> TBL: There are three different scenarios we should distinguish. One, people use flash to make a web site which is still very much hypertext and would be much more reusable if declarative. Two, flash (etc) is used to make a user interface which is not hyopertext at all. Like Slife, tabulator, or maps. there is a concept of identify or place, so URIs are relevant, and indded SL has sluris. Different URI schemes may be appropriate here. Three, the procedural
HT: I would recommend the Sean McGraths keynote cited above.
<ht> Text of one conclusion therein (slide 63): "What we might loose?
<ht> Hypertext and deep linking as we know it.
<ht> Search as we know it (!)
<ht> "Emergent properties: as we have come to know
<ht> them :mashups, folksonomies etc.
<ht> UI simplicity. Grandma won't be able to "surf""
HT: Quotes above from the keynote.
<DanC_lap> (phpht. the KC meeting isn't in the schedule on http://www.w3.org/2001/tag/ . )
SW: Talks about future mtgs after Sept.
HT: f2f East coast early Dec.
<ht> NW: Bad idea -- XML Conference.
SW: Tim, did you say Southhampton Oct/Nov?
<DanC_lap> 23-25 September in Kansas City -- http://www.w3.org/2001/tag/2008/01/31-minutes
SW: Where would we like to meet after Sept?
NW: Proposes January.
NM: Concerned about proximity to TAG election. Could mean that first meeting for new members isn't until several months after they join.
DC: Goes to whiteboard to try and work out dates.
NW: Dec 11,12.
In Cambridge Dec 10,12.
<scribe> ACTION: Stuart to put up Web poll re. dates for Dec 10-12 f2f [recorded in http://www.w3.org/2008/05/21-tagmem-irc]
<trackbot-ng> Created ACTION-157 - Put up Web poll re. dates for Dec 10-12 f2f [on Stuart Williams - due 2008-05-28].
RESOLUTION: to thank the the hosts here at HP and Amanda at the Williams home, with applause.
Break ... resuming at 2:30
<Zakim> timbl, you wanted to suggest different scenarios
Resuming after the break
DO will discuss changes to the document
DO: Added section in Section 2 dealing with failure outcomes.
JAR: Need to discuss outcomes where language is accepted but processor does the wrong thing.
Discussion of wording on section 2
<dorchard> Applications succesfully process texts of an older version of a language.
<dorchard> A newer language is backwards compatible with a newer language if an application written to the newer language successfully process texts of the older language.
<dorchard> A newer language is backwards compatible with an older language if an application written to the newer language successfully process texts of the older language
<Stuart> "A change in the definition of a language is backward compatible if consumers of the evolved language can correctly process text written for the previous version of the language."
Agreement on above wording
The last from Stuart.
NM: Instead of "We specify ...", "For the incompatible strategies there are a range of possible In second to last para encourage people to think about versioning early. 3rd para of 2.1 -- Apps are written to assume a particular version of the text. Often no version id is in the text. If there is an id in the text, and the app supports just one version, then the version id is a crosscheck.
SW: I wish this was all normalized into compatible and incompatible changes to the languages.
<DanC_lap> (I don't think we *need* to define compatibilty in terms of application behavior, but I think talking about that way is straightforward )
AM agrees with DanC
<Norm> I don't actually think we talk about the compatibility of applications
Section 5 -- Text of Ed Note
NW: The second version does not scan.
NM: Against the first one.
<Zakim> Norm, you wanted to observe that intended to be versioned and extensible don't work for me
<Zakim> noah, you wanted to explain why dave's option 1 makes me unhappy
JAR: Perhaps we need a better version of the second alternative
<Norm> Part of the problem here is the scope. I expect most of the readers of this document to be thinking about designing an XML language, not ASCII or XML.
<DanC_lap> noodling... "To facilitate independent evolution of producers and consumers, languages in distributed systems should be extensible".
<DanC_lap> noodling... "Extensible languages facilitate independent evolution of parties in a distributed system"
AM: Norm, this was my comment about the scope of the document
JAR: Echoes Dan's sentiment
DO and NM can live with "To facilitate independent evolution of producers and consumers, languages in distributed systems should be extensible".
DO: Any objections to the above?
<scribe> ACTION: JAR to write up thoughts on versioning and share with the group [recorded in http://www.w3.org/2008/05/21-tagmem-irc]
<trackbot-ng> Created ACTION-158 - Write up thought son versioning and share with the group [on Jonathan Rees - due 2008-05-28].
Vote of Thanks to SW and HP for hosting
<scribe> ACTION: David to update compatibility strategies document in response to f2f discussion [recorded in http://www.w3.org/2008/05/21-tagmem-irc]
<trackbot-ng> Created ACTION-159 - Update compatibility strategies document in response to f2f discussion [on David Orchard - due 2008-05-28].
SW: The Bristol F2F has concluded.