W3C

Technical Architecture Group Teleconference

26 Apr 2012

See also: IRC log

Attendees

Present
Robin Berjon, Jonathan Rees, Henry Thompson, Noah Mendelsohn, Ashok Malhotra, Jeni Tennison, Yves
Regrets
Peter Linss, Larry Masinter
Chair
Noah Mendelsohn
Scribe
Robin Berjon

Contents


NM: probable regrets on the 10th

JT: probable regrets on the 10th from me too

NM: Jeni, can you scribe next week?

JT: yes

Approve minutes of prior meeting(s)

f2f minutes http://www.w3.org/2001/tag/2012/04/02-agenda

NM: Objections?

[none]

HT: I note that there are still a bunch of editorial red marks in my sections
... people haven't gone back and made the necessary changes — none of them are serious
... not objecting to approval

RESOLUTION: Minutes from the f2f are approved

http://www.w3.org/2001/tag/2012/04/12-minutes -> 12/04 minutes

NM: freshly arrived, people can ask for time
... objections?

[none]

RESOLUTION: Minutes from the 12/04 are approved

http://www.w3.org/2001/tag/2012/04/19-minutes -> 19/04 minutes

NM: look good to me

RESOLUTION: Minutes from the 19/04 are approved

Administrativia

NM: I believe that people need more discussion on XML-ER, so it's put to you
... and Robin has asked about election procedures
... hearing no changes to the agenda

<noah> ACTION-687?

<trackbot> ACTION-687 -- Noah Mendelsohn to look for opportunities to discuss putting forward something to the AB about the Process and the failed reference from REC drafts to expired RFCs as a side-effect of scope creep etc. -- due 2012-05-01 -- OPEN

<trackbot> http://www.w3.org/2001/tag/group/track/actions/687

<noah> Proposal e-mail: http://www.w3.org/2001/tag/2012/04/19-minutes

NM: seemed convoluted, sent email, made a proposal based on responses

<noah> Proposal e-mail: http://lists.w3.org/Archives/Public/www-tag/2012Apr/0185.html

NM: can send a note to the AB without further discussion

<noah> ACTION-687?

<trackbot> ACTION-687 -- Noah Mendelsohn to look for opportunities to discuss putting forward something to the AB about the Process and the failed reference from REC drafts to expired RFCs as a side-effect of scope creep etc. -- due 2012-05-01 -- OPEN

<trackbot> http://www.w3.org/2001/tag/group/track/actions/687

JAR: believe further iteration is needed. The iteration might lead to a decision to do nothing, that would be ok

NM: would like to handle this in email

NM: some time ago the TAG agreed that the work on HTML Data had been successfully completed
... I was tasked with recording that in the product page

<noah> http://lists.w3.org/Archives/Public/www-tag/2012Apr/0148.html

<noah> On 18 January: http://www.w3.org/2001/tag/2012/01/19-minutes.html#item05

<noah> <noah> RESOLUTION: The draft product page at http://www.w3.org/2001/tag/products/htmldata.html is agreed as the basis on which the TAG closes out it's work on Microdata/RDFa coordination

NM: this email points out that on 20120119 we resolved the above
... my view was the TAG passed a resolution, I took an action, announced it, and propose to close
... but today, LM emailed about it

<noah> Larry asks to take this to Rec: http://lists.w3.org/Archives/Public/www-tag/2012Apr/0209.html

JT, AM: Robin pushed back

<JeniT> http://lists.w3.org/Archives/Public/www-tag/2012Apr/0209.html

RB: push back was on XML, not HTML Data

YL: I think it would be difficult for the TAG to have the cycles to move everything to REC
... we know that there's a good start in both cases
... it's fine for the TAG to say it did its share
... without necessarily push to REC
... pushing these documents to REC can be done later, I think that closing the action and the product is in order

JT: in the HTML Data work there were two notes produced with the intent that they could be turned into something more solid
... especially the microdata to RDF conversion

NM: TAG needs to be involved?

JT: not necessarily directly, but W3C needs to find a good home for it

NM: action to check up on whether W3C is doing the right thing there, possibly in a few months?

close ACTION-664

<trackbot> ACTION-664 Announce completion of TAG work on Microdata/RDFa as recorded in http://www.w3.org/2001/tag/products/htmldata.html and to finalize the product page and associated links closed

<scribe> ACTION: Jeni to check that W3C has found a good home for the output of the HTML Data TF, especially microdata/RDF conversion - due 2012-10-26 [recorded in http://www.w3.org/2012/04/26-tagmem-minutes.html#action01]

<trackbot> Created ACTION-699 - check that W3C has found a good home for the output of the HTML Data TF, especially microdata/RDF conversion [on Jeni Tennison - due 2012-10-26].

NM: I think that's an appropriate resolution in the particular case of Microdata/RDFa. If Larry (or anyone) wants to ask the TAG to consider whether, in general, more of our work should be REC-track, that would be a separate discussion for them to request.

XML-ER

ACTION-656?

<trackbot> ACTION-656 -- Noah Mendelsohn to schedule discussion of possibly getting W3C to invest in technologies for liberal XML processing (e.g. XML5) -- due 2012-04-24 -- PENDINGREVIEW

<trackbot> http://www.w3.org/2001/tag/group/track/actions/656

NM: JT framed the proposal

<JeniT> http://lists.w3.org/Archives/Public/www-tag/2012Apr/0169.html

NM: LM specifically asked that the TAG's work on HTML/XML should go on the Rec track
... would like not to discuss that now, we will see Norm in June, and can discuss in preparation for that

NM: would like to focus on XML-ER CG, goals, use cases, etc.

JT: HT asked me to go through the minutes from f2f and pull out areas that we had raised as concerns
... put those in email
... I think that we should engage positively with the XML-ER CG

JT: looking perhaps to drop them an email suggesting changes in their charter

<noah> http://www.w3.org/community/xml-er/wiki/Charter

JT: focusing on what we would say if we spoke to them
... concern from the minutes are listed in my email, refer to that

<noah> Concerns raised by TAG members during the F2F discussion included:

<noah> * restricting XML-ER processing to non-safety-critical applications

<noah> * ensuring that any error recovery is reported noisily

<noah> * error recovery causing a race to the bottom and evolutionary drift

<noah> * potential security problems with the same file being interpreted in different ways by different processors

<noah> * interactions with media-type sniffing

JT: if I were to communicate with them, I would need help to provide more detail on some of the concerns

NM: some question in my mind as to what the level of interest the TAG has in dealing with this

<jar> XML-ER if it exists should have its own media type

NM: fine for me to dive in, but want to make sure that people are really interested
... we don't owe it to anybody to do more

<noah> RB: Would it be simpler if people would bring concerns directly to the community group?

<JeniT> +1

YL: some concerns in JT's email are already in the charter

<jar> "Backwards compatible with XML 1.0." requires error reporting

YL: critical apps would simply reject ER, backwards compat is taken into account

YL: I agree with RB that if there are specific issues they can be taken directly to the CG

HT: I think that this is close enough to a number of essential architectural issues that I don't want to leave it to just CG discussiojn
... we should discuss this as the TAG

<noah> Henry, can you give an example of something the tag >might< want to say?

HT: I'm sufficiently concerned about this at the architectural level that I want to keep it on our agenda
... I'm not saying that we should be tossing bombs over the parapet to them

<noah> To motivate your "outlier" view that we keep it on the table

JAR: I agree with that, it seems that we've been talking about extension points and the such for years and we're closer to that now

NM: some in the group seem to think we can just interact with the CG
... henry would like to keep it
... JAR thinks it's useful to discuss

NM: HT do you have examples of TAG level concern

HT: several points in the discussion where JT|RB said "we agree, I expect it will turn out that way"
... but if it doesn't, we have a problem
... I would like to capture and ensure those

NM: YL asserted that processing critical applications is covered by the charter
... I don't think that's the only way of looking at it
... the scope is set, but if software is confused it will have a flag

<JeniT> it's w community group, not a working group

<noah> NM: Yves makes the case that, because the charter mandates a warning on fixed up output, we're OK on the "critical apps" front. Not necessarily. There's still reason to question whether the charter should have mandated a style of fixup that would have been suitable for a broader range of applications...

<noah> NM: Of course, Anne's done a wonderful service by moving ahead to meet what he (and others) see as the goals, and we'd lose that if the goals changed a lot.

YL: first I wanted to reply to HT that having people contributing to the CG directly is not incompatible with finding issues and working on those
... I think it will be faster if people comment directly to the CG
... 2nd point is that it's a CG, it's not tasked to produce a Rec, I wouldn't worry too much about small details

<jar> wiki has no pointer to mailing list

<JeniT> jar, the home page for the CG has the link on the left

<noah> I'm not saying what the WG is doing is wrong or bad. I'm saying that the goals weren't debated as broadly as we do for some other work.

YL: in the charter and such — I think the fact they added that errors are surfaced at the application level is a sign that they want to tackle applications possibly rejecting content

<noah> In practice, going down this path is probably the right thing for now.

YL: taking into account security-critical applications
... but I thikn it's a good indication, and we can trust the process of the CG
... and monitor it

<jar> http://www.w3.org/community/xml-er/

<noah> Right Robin...but the point you're not addressing is that the fixups themselves are designed for interactive browser applications.

<noah> RB: I think it will be faster to bring concerns to the CG directly. It's a CG, not a WG. Doesn't formally need a charter. That was done to be a helpful point of reference.

<noah> RB: It's not aimed at "error recovery" it's designed to take any input and produce a parse. Not sure the concerns about critical apps apply

<noah> I note that the group is titled XML-ER

<noah> RB: XML-ER naming is the result of my bad joke, now regretted.

<jar> http://lists.w3.org/Archives/Public/public-xml-er/

HT: maybe RB did just say so, but thanks for reminding us that this is not a WG which changes the dynamic
... the charter is just a convenience and isn't binding
... but it's an indication of direction
... we may be headed towards a situation in which apps can opt-out of ER
... but I'm not sure I want it that way, I think I want it opt-in
... nobody is ever going to see fixed up output unless they take steps to
... it shouldn't be the default

NM: isn't that somethign that different processors e.g. browsers might default differently
... ?

HT: I don't think so, but we'll have to see how it develops

AM: really any inpuyt?

<jar> This is very interesting… very similar to sniffing

RB: yes

HT: he did, which is reasonable so long as it's deterministic

NM: this is similar to HTML5 where it does that
... this can include some complex parsers for HTML
... but I don't think that this is reasonable for e.g. importing to a DB
... but you can imagine that some fixups are low-risk
... e.g. upper/lowercase

JAR: that doesn't sound good for XML

NM: right

JAR: XML assigns errors to some strings
... this is incompatible with XML

NM: this will operate successfully with a lot of apps that expect XML
... we're talking about when this is appropriate
... do I ever want to import broken XML to XML tooling?

JAR: this is exactly the same question about authoritative metadata and sniffing

NM: there's a move in teh community that XML is not successful on the Web because it is too strict
... XML-ER builds a tree for "broken" content

JAR: not arguing the merits, the TAG has been here several times
... why would we say something different?

NM: the community is asserting that XML, which is important to W3C, is having far more limited impact than we wanted
... trying to be helpful to a broader range of things that people are doing
... without crashing airplanes

JAR: just saying that we shouldn't take this in isolation, should use the context of authoritative metadata

HT: JAR's question made me realise that I'd like to hear how this sits with the notion of media type
... as JAR pointed out, the XML spec says that a string of characters which doesn't satisfy the condition for WF
... is not XML, it's just characters
... it's not XML with errors
... delicate but relevant point
... people would be comfortable with saying "this is Fortran with a bug", but people don't say that about XML

NM: what usually happens is that for programming languages, the spec is strict but they can resync

HT: I deny that — they define sync points so that the compiler can give errors
... main point is where does this fit in the space that we know about in terms of media types
... content type but also accept headers
... unlike text/html which is being redefined, the jury's still out on what they say
... but they might say that any content might legitimately be served as text/html
... several people have made clear that they hope the goal of the XML-ER is not to redefine the application/xml media type

<noah> I hope they don't say that any content is validly served as text/html. I hope/expect they will make a massive application of Postel's law, and say legally served content MUST validate, but clients may be liberal in what they process.

<noah> RB: I think the media type question is very much open in the CG.

<noah> I think Henry was talking about the likely registration of text/html

<noah> Not anything to do with the CG

<noah> RB: The question was how to make XML usable in various situations without breaking things.

<noah> RB: Nobody has yet looked in detail at whether to recommend use of application/xml, which would be a significant change the registrarion

NM: also a question about whether text/html will sets a precedent

HT: we're still waiting on that one, but we'll have to look at it

JAR: regardless of what the CG decides to do, this is a very interesting question, I see parallels with other issues
... we should keep this going

<ht> We need a Postel's Law issue

JAR: maybe we should wait until someone has something to say about it
... but shouldn't close

+1 on a Postel issue

<jar> +1

<noah> ACTION-696?

<trackbot> ACTION-696 -- Jeni Tennison to frame discussion of XML-ER goals and use cases -- due 2012-04-24 -- PENDINGREVIEW

<trackbot> http://www.w3.org/2001/tag/group/track/actions/696

<noah> ACTION-656?

<trackbot> ACTION-656 -- Noah Mendelsohn to schedule discussion of possibly getting W3C to invest in technologies for liberal XML processing (e.g. XML5) -- due 2012-04-24 -- PENDINGREVIEW

<trackbot> http://www.w3.org/2001/tag/group/track/actions/656

NM: this action dates from before the CG, my work is done
... close both?
... to keep this on the table, what's the next step?

JAR: someone to think about this
... I see big parallels with httpRange-14

NM: I was hoping you wouldn't say that

HT: I agree with JAR, and agree it's going to be hard to find something to say about this

<jar> issue-20?

<trackbot> ISSUE-20 -- What should specifications say about error handling? -- open

<trackbot> http://www.w3.org/2001/tag/group/track/issues/20

HT: we have an issue similar to hr14 which keeps coming up: is Postel's Law of any use?
... if not, should we write an obit?
... if it is, can we state so?
... if we have something different what's the delta?

NM: I'm not sure that's as fraught as hr14

Scribe notes famous last words

NM: seems close to authoritative metadata indeed
... Postel's law is out there to advocate in favour of robustness over safety
... trying to get to the ongoing effort about HTML/XML unification
... tempting for me to say that the bits that are specifically about XML should go to that TF, and anyone is welcome to do that
... HT is saying that we could invest in the deeper quesiotn of Postel's Law and its relationship with authoritative metadata
... anyone want to do the work?

HT: want to yes, but can is a different question

<jar> ditto

NM: this is significant if done well, but we need commitment

<noah> close ACTION-696

<trackbot> ACTION-696 frame discussion of XML-ER goals and use cases closed

<noah> close ACTION-656

<trackbot> ACTION-656 Schedule discussion of possibly getting W3C to invest in technologies for liberal XML processing (e.g. XML5) closed

NM: if someone wants to bring this up again, I'll be sympathetic so long as they can point out what's changed

<noah> NM: To sum up, the XML-specific part of this may come up again in the context of the HTML/XML unification effort, which is ongoing.

<noah> NM: Otherwise, asking to reopen focus on XML-ER is in order >if< someone steps up to move it forward and do real useful work on it.

<jar> error handling and extension points are very closely related

<noah> NM: Likewise, starting a major effort on the tension between authoritative metadata and Postel's law sounds very cool ( to the chair anyway ), but only if someone is ready to do months of work on it.

and versioning!

TAG Election Procedures

NM: framing from the chair
... number of emails flying in various quarters about changing the TAG and all that
... before Sophia I asked if we wanted to talk about that, but it was rejected

HT: for discussion at the f2f

NM: it may be better to talk of this f2f though, can have lunch discussions and the such
... first of all, it's been noted several times that changes to the process are not things that we drive
... but we can ask for them
... received objections to having this discussion at all
... so for this afternoon, the scope is strictly about election procedures
... if people have other suggestions, please send them in email
... these are time-consuming so please set the bar high
... I get nervous when we get too far in proposals for change without being clear about what we are trying to change

<JeniT> http://lists.w3.org/Archives/Public/www-tag/2012Apr/0105.html

<noah> RB: I sent some feedback to the member list, felt encouraged by the response, so wanted to bring it forward for wider discussion. Two of these are just to practice, vote counting is a change to process.

<noah> RB: E.g. to vote counting

<noah> (Hmm...I thought the counting procedure was at least implicitly part of the process)

<noah> RB: Ideas:

<noah> RB: 1) Make nominee list public (don't think it is)

<noah> RB: 2) Avoid tactical voting, probably as embodied in WBS (to avoid tactical voting...perceive that members avoid casting second vote when first choice is at risk)

<noah> RB: 3) Have a public mailing list on which people can discuss the election with the candidates, get answers from the candidates.

<Ashok> I think the recommendation is to use preferential voting -- i.e. first, second, etc.

HT: I strongly endorse the change to Process to avoid tactical voting
... I'm conscious that it's awkward to say so
... but I will say that in every election I have stood in I have voted only for myself
... and I think that's broken

s/but I will say that in every election I have voted in I have voted only for myself/but I will say that in every election in which I stood I have voted only for myself

JAR: I think that we need to look at the broader problem and wonder if election reform will solve that
... the problem is that we want abilities we don't have
... I don't think that this solves that

<ht> Oh yes, and I meant to say contra LM in email that it's precisely when the number of candidates is just larger than the number of seats that tactical voting is most tempting

NM: I don't want to pull in the entire scope of changing the TAG

<jar> the problem is getting constituencies represented, and getting expertise in areas where we're weak

<JeniT> RB: I've spoken to people who have wanted to run, but didn't bother because they didn't feel they had a chance of winning

<noah> RB: Don't focus just on counting. Right now, people who aren't well known in the AC don't run, because they perceive that without name recognition in the AC they can't win

NM: two or three separate things that may be in contradiction
... one is that I think that RB is making good points in isolation
... tactical voting bad, people telling their story good
... two, be careful. If you look at who's running, there are some issues that aren't being discussed here
... as chair I feel tension between what we need to deliver and the notion that people put themselves to run
... but ACs don't ask if people can write
... but writing skills are really important for the TAG
... three, the TAG is a funny group
... I have an opinion about it, but others see it differently
... see its goal as making people happy
... but it seems that if you're going to do more than very small fixes to the process then you're going to have to look at broader questions
... one point of view is lets at least fix the small things, put the bigger things on the table later
... but there's the risk that people will perceive that we're fixing the bigger issues
... one thing I will fight against is backing into revisiting what the TAG is about
... it's important, but it's something that needs to be done with care
... to some degree the TAG was chartered in part to be unpopular
... and look at inconvenient things
... it's really hard for me as chair to know when we're doing our job and when we're just being stupid

<jar> RB, do you agree with what I said (that process changes are a means to an end), do you agree with what I said the end was, and how far do you think the process changes go toward achieving that end, 10%, 50% 90%?

<noah> NM: You mean the particular 3 changes you proposed.

<noah> JAR: Yes.

<noah> RB: What means to what ends?

<noah> JAR: The one I said.

<noah> JAR: Bringing better constituency representation and more expertise.

<noah> RB: That's what I meant by better candidates

<noah> RB: Chances of success are hard to judge. One "better" person out of 5 might be good.

NM: there are TAG members who in retrospect turn out to be stronger and that's great
... but at times we need several, it may be better to have several people on one topic at times
... no corporation would appoint us in the way we are
... I think Tim's appointees are often the strongest

<jar> RB ventured 40%, I think… I'm satisfied with that kind of answer, but note that in future we need to talk about the other 60%

NM: and I think that he uses his vision for that
... I'm not convinced that the AC takes that into account
... the time investment is pretty significant
... it's good that independents are willing to stretch
... but it's hard without deep corporate pockets
... if you're willing the grant that there were problems implicitly solved in RB's proposal
... I think there's agreement that these are small steps in the right direction
... but should the TAG do something with this?
... individuals can go to the AB directly
... TAG aware of issues, point out sympathy on the TAG for solving this
... point the AB to these minutes

<Yves> if the TAG says "yes it should be fixed in a way" it would be a good indication that individual claims are valid

JAR: I see LM's point that the TAG doesn't do process
... but that's not the end of the story
... in order for the TAG to address its charter the TAG needs specific people
... RB's proposal is about helping with that

NM: it would take weeks and months for the TAG to discuss the broader issues
... but it would take months and we haven't done it yet
... trying to suggest that people here approach Team and AB pointing to these minutes
... it does not the question about are we staffing the TAG right

<jar> if it did 40% that would be huge

NM: if it's the only change we make in ten years, I don't want it to happen
... but if it's a small tweak we can do without any presumption that no further changes and debate will happen, then it could be taken to the AB/Team

RB: happy to go to the AB and point to these minutes

NM: if you want, draft a note, send it to the member list, and give us a chance to review
... and let me as chair draft another note giving context and larger issues
... capture informal feeling that there are concerns

<scribe> ACTION: Robin to send note to tag@ that he will send later to the AB (as himself) proposing the changes to electoral proceedings [recorded in http://www.w3.org/2012/04/26-tagmem-minutes.html#action02]

<trackbot> Created ACTION-700 - Send note to tag@ that he will send later to the AB (as himself) proposing the changes to electoral proceedings [on Robin Berjon - due 2012-05-03].

<scribe> ACTION: Noah to follow up with Robin on election reform [recorded in http://www.w3.org/2012/04/26-tagmem-minutes.html#action03]

<trackbot> Created ACTION-701 - Follow up with Robin on election reform [on Noah Mendelsohn - due 2012-05-03].

<noah> ACTION: Noah to follow up with Robin on election reform proposals [recorded in http://www.w3.org/2012/04/26-tagmem-minutes.html#action04]

<trackbot> Created ACTION-702 - Follow up with Robin on election reform proposals [on Noah Mendelsohn - due 2012-05-03].

action-702 closed

<trackbot> ACTION-702 Follow up with Robin on election reform proposals closed

NM: remind me of what you'd like discussed

[adjourned]

trackbot, end meeting

Summary of Action Items

[NEW] ACTION: Jeni to check that W3C has found a good home for the output of the HTML Data TF, especially microdata/RDF conversion - due 2012-10-26 [recorded in http://www.w3.org/2012/04/26-tagmem-minutes.html#action01]
[NEW] ACTION: Noah to follow up with Robin on election reform [recorded in http://www.w3.org/2012/04/26-tagmem-minutes.html#action03]
[NEW] ACTION: Noah to follow up with Robin on election reform proposals [recorded in http://www.w3.org/2012/04/26-tagmem-minutes.html#action04]
[NEW] ACTION: Robin to send note to tag@ that he will send later to the AB (as himself) proposing the changes to electoral proceedings [recorded in http://www.w3.org/2012/04/26-tagmem-minutes.html#action02]
 
[End of minutes]