Re: {minutes} 9 December 2010 SVG WG telcon

On Thursday, December 9, 2010, 10:40:04 PM, Adrian wrote:


(Adrian, the tracker system need a plain-text copy of the minutes, in the body of the email and not as an attachment, to be able to auto link to issues and actions).

                   SVG Working Group Teleconference

09 Dec 2010



   See also: [3]IRC log



          heycam, adrianba, pdengler, ed, +39.537.7.aaaa, Shepazu,

          Erik Dahlstrom

          Adrian Bateman


     * [4]Topics
         1. [5]SVG 1.1F2 progress
         2. [6]Test Suite
         3. [7]SVG elements in HTML <head>
     * [8]Summary of Action Items

   <trackbot> Date: 09 December 2010

   <scribe> ScribeNick: adrianba

   <scribe> Scribe: Adrian Bateman

   <scribe> Agenda:


   ed: agenda is short, agreed on the last telcon to focus on 1.1
   ... need to see what progress we've made

SVG 1.1F2 progress



   ed: if we could go through the list from top to bottom

   <ed> ISSUE-2339?

   <trackbot> ISSUE-2339 -- Last Call Comment: definition of azimuth,
   elevation for feDistantLight -- open

   <trackbot> [11]



   <trackbot> ISSUE-2339 -- Last Call Comment: definition of azimuth,
   elevation for feDistantLight -- open

   <trackbot> [12]


   heycam: anthony said he'd get to this later today

   ed: think this is one that i think i wrote tests for but not sure
   ... might be one we can put in and not have a test ready for initial
   testing phase
   ... okay, so progress on that one


   <trackbot> ISSUE-2334 -- Last Call Comment: filter primitive
   subregion and feGaussianBlur, feTile and infinite filter input
   images -- raised

   <trackbot> [13]


   ed: this one i've been working on a bit today, i commited some
   changes for the second half to clarify
   ... the first part is not yet addressed but we have an agreement on
   how it will be resolved and i started with some tests to see what
   implementations are doing at the moment
   ... this is one i think we will need tests for
   ... i don't want to put something in the spec before i have
   something ready in tests

   heycam: so you'll be able to do this tomorrow?

   ed: yes, that's my plan


   <trackbot> ISSUE-2335 -- Last Call Comment: Clarify feConvolveMatrix
   bias property -- raised

   <trackbot> [14]


   ed: i had an action and so did anthony - my action is closed because
   i think anthony's to put in the wording, he has been following up on
   this to get agreement
   ... i saw a new test was added on this

   <scribe> ACTION: ed to review the test for ISSUE-2335 [recorded in

   <trackbot> Created ACTION-2919 - Review the test for ISSUE-2335 [on
   Erik Dahlström - due 2010-12-16].

   ed: not sure if this is the one anthony said he commited changes to
   the spec
   ... yeah, think this is one of them


   <trackbot> ISSUE-2338 -- Last Call Comment: type of feFunc* --

   <trackbot> [16]


   tav: it's done

   ed: marked that one as being done in tracker

   heycam: does it have all the comments for disposition of comments?

   ed: yes, when we do the extraction of tracker it will be okay
   ... but we need to go through the remaining issues and ensure they
   have the notes but we can leave that for now

   heycam: does this need to be done in time for the publication?

   chrisl: that needs to be done before the transition request?

   heycam: does the transition involve a phone call?

   chrisl: yes

   heycam: we haven't planned for that and if the pub deadline is
   wednesday we need to that before

   chrisl: we need to show we've exited LC and CR - could be one
   meeting to do both

   shepazu: that'll be team contact and maybe the chairs and the
   director and domain lead
   ... probably plh, ralph and either chris and/or doug
   ... probably won't be able to schedule at this stage

   chrisl: we need to go to the chairs and there needs to be notice for
   people to object if they feel comments not dealt with

   heycam: that's unfortunate

   shepazu: we won't be able to publish but we can do all the other

   heycam: okay

   shepazu: so we'll publish in early january

   <ChrisL> 31 January, in answer to the question

   ed: think we should aim for both disposition of comments and
   implementation report before christmas then



   heycam: think that's feasible because working hard we would have met
   the wednesday deadline

   ed: aiming for something earlier helps to get done

   heycam: if we have an extra couple of weeks then we might consider
   fixing other things in the tests

   chrisl: would much rather see people running this sooner so we have
   a feel for where we are
   ... we don't know how many tests we have where no one passes yet


   <trackbot> ISSUE-2343 -- Last Call Comment: 15.12 Filter primitive
   ‘feComposite’ formula -- raised

   <trackbot> [18]


   <ChrisL> we can easily run spot rechecks as needed for tests where
   we find bugs

   ed: this one was done - not sure if the issue itself was updated
   ... yeah, we still need approval of response

   heycam: is there a minimum amount of time we need to give commenters
   to respond?

   <ChrisL> well, a reasonable time. some days typically

   ed: i don't think the remaining issues are that hard to resolve
   ... not sure if new tests are needed - we can work this out in the
   coming days
   ... we'll ask anthony if there are new tests needed, seems mostly


   <trackbot> ISSUE-2346 -- Last Call Comment: previous discussion
   about filterRes -- raised

   <trackbot> [19]


   ed: i've closed this


   <trackbot> ISSUE-2351 -- Last Call Comment: Clarify that units are
   required on <length>s in style attribute; fix examples in text. --

   <trackbot> [20]


   ed: i didn't close the issue - not 100% if done

   tav: yes, it was done

   ed: okay, i'll close then
   ... we should keep issues open so we can more easily extract them


   <trackbot> ISSUE-2364 -- Last Call Comment: SVG 1.1 may be ambiguous
   about the root element acting as a proximal event target -- raised

   <trackbot> [21]


   shepazu: i'm working on this one - apparently svg doesn't use the
   term hit-testing so looking at a definition for that
   ... not done yet but i expect to have submitted it to cvs today and
   be ready for review

   heycam: who's comment was it?

   shepazu: there were several - not sure if it was a last call issue -
   don't think it was
   ... came about because of discussion after last call about whether
   the svg root should intercept pointer events
   ... two different contexts, inline or embedded by reference

   <ChrisL> Cam, see [22]
   for the modified disco.xsl and some sparse documentation in


   heycam: so what were the decisions?

   shepazu: the interaction section mixes a bunch of things including
   action processing order and we decided that we were going to clarify
   the section on ui events

   heycam: not sure if we have tests for this

   shepazu: we probably don't but not sure if the changes i am making
   will result in tests, probably but i haven't looked to see if there
   are tests yet
   ... i need to make sure there are tests that do check this

   heycam: there are 3 tests that may be testing something around here


   <trackbot> ISSUE-2368 -- Problems with grammars for numbers --

   <trackbot> [23]




   heycam: i couldn't see who was the original commentor - the issue
   was that the grammar for the points production for polygon and
   polyline; if you don't use a . in the number you are required to use
   an exponent
   ... think it's because floating point and integers are separate and
   if you only look at the fp production then you might think integers
   are missed out
   ... but if you take them both together then it's fine
   ... waiting for feedback


   <trackbot> ISSUE-2384 -- Order of rx / ry computation for rounded
   rects -- raised

   <trackbot> [25]


   pdengler: i think the clarification makes sense - it's ambiguous
   without it
   ... we should update the tests but we don't know if we'll be able to
   fix it on our side

   heycam: do you agree with the proposed fix but don't have time to

   pdengler: yes

   heycam: will you be able to fix in the future?

   pdengler: we'd want to get to it, yes

   heycam: think it's an edge case

   ed: yeah, it won't be noticeable to most people and it's possible to
   work around

   pdengler: true

   <scribe> ACTION: heycam to add the wording and fix the test for
   ISSUE-2384 [recorded in

   <trackbot> Created ACTION-2920 - Add the wording and fix the test
   for ISSUE-2384 [on Cameron McCormack - due 2010-12-16].


   <trackbot> ISSUE-2391 -- Last Call Comment: better changes appendix
   -- raised

   <trackbot> [27]


   chrisl: going through all the logs to find all the changes since
   last publication is taking time

   heycam: is this something you're part way through?

   chrisl: not started yet, don't know if anyone has updated the
   appendix and should be able to start again

   pdengler: when we last looked it hadn't been updated

   chrisl: have to go through the CVS change log and read the diff and
   summarise the change
   ... yes, you can pick the starting point
   ... we got a comment saying the changes appendix needs to be better,
   for the patent policy we need the list of changes from the last time
   we published
   ... because any new features added after LC doesn't get covered
   ... so we need to show no new features were added and just
   ... that part is fine but the 'better' appendix is more work
   ... the current version is less detailed than what was asked for in
   the comment
   ... to turn it into a reference for someone that knows the orginal
   edition and want to just know what has changed
   ... we pushed it back to the commenter
   ... but they haven't got to it

   ed: i dealt with the other two remaining issues
   ... is there anything else we need to do with the spec that isn't
   listed here?

   heycam: i remember noticing at one stage that the style for the dom
   method summaries was looking good in some browsers but not in others
   ... i want to make sure it looks alright in most browsers - just a
   one off quick check at the end when we're ready

   ed: just trying to make sure we don't have any surprises, that's all

Test Suite



   ed: tried to keep track of issues on this wiki page



   ed: one thing to note is that i have only kept this one updated and
   only contains the tests we have accepted
   ... the other harnesses include draft tests including some not
   ... this is the official one for the implementation reports

   heycam: does this mean we won't have new accepted tests?

   ed: i think we have to be open to accept some new ones or possibly
   dropping tests back to draft
   ... don't think it's possible to say until we've done the testing
   and see where we are
   ... it may show issues we've missed - i think this harness and the
   tests are in good shape
   ... and it's useful to run the tests and learn about any issues left
   to fix

   heycam: i think there are only a couple that need changes
   ... 99% of the tests are fine regardless of any ambiguities in the
   pass criteria

   <ChrisL> Updated the disposition of comments -


   heycam: so i agree we can do the run through and if there are
   substantial changes to the tests we can run those individually

   chrisl: there was a discussion of bidi include testing arabic - was
   testing two things
   ... so i made it like the 1.2 test that was simpler



   heycam: because i wasn't sure if the test required a particular font
   i couldn't tell if it was a font problem

   chrisl: the reference images should be good references with the
   right font

   ed: regarding bidi and text anchor we made clarifications in 1.2 and
   we need to bring this back so the specs are compatible

   chrisl: 1.1 1st edition had some well meaning but incorrect text

   ed: chrisl, will you do the back porting?

   chrisl: i could look at this, do you know where it is?

   ed: i sent mail to the list with the change list - it's multiple
   changes but should be able to figure it out

   <scribe> ACTION: chrisl to bring the 1.2 changes for bidi and text
   anchor back to the 1.1 spec [recorded in

   <trackbot> Created ACTION-2921 - Bring the 1.2 changes for bidi and
   text anchor back to the 1.1 spec [on Chris Lilley - due 2010-12-16].

   heycam: as part of this will you look at the tests?

   chrisl: i started this by simplifying the tests, made some new tests
   that used a WOFF font (unapproved)
   ... it's there so eventually we can have better tests without having
   to download and install fonts
   ... -06, -11, and -12
   ... -06 was the only approved one, the others won't be in the
   implementation report - just so we remember

   heycam: does -06 require woff?

   chrisl: no, it still uses platform font, just simplifies by only
   looking for mandatory ligature

   pdengler: on textintro-01, i got lost looking at the test on the
   ... is it that the reference images haven't been regenerated?

   chrisl: this is -06 we're talking about
   ... i've checked in a ref image sitting in the image patches
   directory but someone is going to have to run the build scripts for
   the harness



   heycam: comparing -01 and -04, not sure why there is combining in
   one and not the other

   <scribe> ACTION: chrisl to fix the text-intro-04 test with feedback
   that has been around a little while [recorded in

   <trackbot> Created ACTION-2922 - Fix the text-intro-04 test with
   feedback that has been around a little while [on Chris Lilley - due

   <ChrisL> ARJ> In the Yiddish example, U+05D0 HEBREW LETTER ALEF is
   followed by U+05B8

   <ChrisL> ARJ> HEBREW POINT QAMATS. The qamats is a non-spacing mark
   and should be

   <ChrisL> ARJ> positioned below the alef, yet in the reference image,
   it is positioned on

   <ChrisL> ARJ> the left side.

   heycam: it takes a long time to run through the tests - it's hard to
   see sometimes if there is a pass

   chrisl: yes, sometimes you have to reload several times, which might
   have been a mistake

   <ChrisL> the ones with multiple animations would be better split
   into separate tests, if you have to hit reload anyway

   <general discussion about lessons learned and test first

   ed: in order to run the test suite, do we need to generate the xml
   ... anthony generated the reporting files, maybe i can rerun the
   ... the problem is last time it included some draft status tests and
   i'm not sure if they have been updated
   ... i can look tomorrow - probably not that hard to regenerate the
   status file
   ... makes sense to see which revision we test so we know if we have
   to go back

   heycam: do we have a general rule if the revision number doesn't
   matter that's okay

   ed: yes, the revision numbers are not part of the tests

   <scribe> ACTION: ed to regenerate the test status file and make sure
   it only contains accepted tests [recorded in

   <trackbot> Created ACTION-2923 - Regenerate the test status file and
   make sure it only contains accepted tests [on Erik Dahlström - due

   ed: i will send mail to the list once the status file is done

   heycam: with animate-element-77, i was asking about whitespace
   causing things to be in the wrong, firefox is wrong but the test
   isn't trying to test whitespace

   ed: if we can leave the test as is for now that is better, i would
   like to have more tests in future for whitespace

   heycam: okay
   ... the other ones i added are just tweaking the pass criteria to
   make it clearer

   ed: when i went through the suite i saw about 20 tests that still
   had red in - better than we were
   ... think it's fine to start running through the suite, if there are
   any issues left we may need to make some more changes



   chrisl: seems to be using Arial - not sure if the script is
   expecting the metrics of Arial

   ed: tried to avoid using Arial where possible but there are some
   cases where it is needed
   ... this is a draft test

   chrisl: i could respond and say it was a draft, but it was raised as
   a comment
   ... okay, it's a draft, i'll leave that one

   heycam: just want to ask about the woff conversion again

   chrisl: i've done some more and checked them in, still some to do,
   some of the tests i changed the font because it was easier than
   converting the svg font
   ... still have three left to do, the rest are done now

   heycam: only a few tests that rely on fonts that haven't been
   ... so patrick, when you run through if there are only a few it
   should be okay to go back to those

   pdengler: yes
   ... we're still on track for giving results next week

   chrisl: what implementation are you running - is it public or

   pdengler: it will be internal so we need to figure out how to get it
   to you

   chrisl: need to get a version to me under member/team
   confidentiality for spot checking

   pdengler: yes, we'll need to do this - just may take time
   ... for clarify, my test team monitors the working group mailing
   list so as soon as it is sent out that it is ready to go we'll start

   heycam: so you'll run all the tests, create the implementation
   report, and then coordinate with chris?

   pdengler: we'll run it as soon as we can and after that we'll get a
   build to chris

   heycam: today after i've done the little issues, i'll mail out and
   say the tests are ready to run

   pdengler: suspect we'll hit some problems and a few will need to be
   re-run anyway

   ed: only thing missing is the xml file which is on me to figure out,
   will get it tomorrow morning

   heycam: don't need it to start testing

   ed: no, but good to get it done for later

   heycam: sounds good

   shepazu: another issue from whatwg that i'd like to briefly discuss



SVG elements in HTML <head>

   shepazu: basically someone was doing svg and when tried to validate
   it then when put svg in body he could reference with use elements in
   other parts of html
   ... but if he put the svg inside the html it used up space and in
   the head it wouldn't validate

   pdengler: why did it take up space?

   heycam: because it's default CSS

   chrisl: display:none should support this?

   shepazu: could also say width height zero

   chrisl: that might affect the viewport

   shepazu: this is maybe stuff for the integration spec - i think
   we're going to have people that expect to be able to put things in
   the head that doesn't display unless you use it elsewhere

   chrisl: we should push this back to the html wg and say it always
   has display:none when it goes in the head

   ed: don't want to do subsetting

   shepazu: no, we don't want to subset, any svg element should be
   allowed in the head
   ... should just be svg allowed in the head as chris said

   pdengler: feels like we're creating work here

   shepazu: no, i think this should fall out of the model - there is a
   quirk of the html spec that doesn't allow svg in the head

   heycam: it's consistent with that you can't have things like
   paragraph in the head

   chrisl: but html doesn't have the same reference model

   pdengler: i think we should figure out a better model before we go
   an push on this - we should think about the bigger picture before
   making requests

   shepazu: i disagree, i think you should be able to put it anywhere
   in the doc and use it anywhere

   ed: and it works but just doesn't validate

   shepazu: think it's not an architectural argument

   pdengler: i think it goes to the fact there are concepts in svg that
   may make sense in html

   shepazu: i think it's orthogonal to that - we can have a
   conversation about broader issues
   ... maybe html will eventually have a use element for re-use but
   that is orthogonal for whether svg is allowed in the head

   pdengler: if you introduce this into html then wouldn't you have to
   make bigger changes to html and what can go in the head?

   shepazu: i'd like to e-mail the html working group

   ed: sounds like we don't have consensus yet but we could start the

   <discussion about whether html might have a use-like thing>

   chrisl: html might add something but we already have one in svg
   ... so we should deal with the svg issue now

   shepazu: i think this will be a common pattern - lots of people
   think this way

   ed: just because something doesn't validate doesn't mean it isn't

   shepazu: i don't think the validation failure makes sense
   ... if in the future people decide to add something like this in
   html and the consequences in the head then that's a bigger issue

   heycam: i hear patrick saying that we should be taking the wider
   issues into account so we don't make it harder in future

   shepazu: i understand but i think they're orthogonal

   heycam: so that's the disagreement

   pdengler: how about we test the waters and see what they think?

   shepazu: i'm fine with approaching the html wg and saying we see two
   ways to solve this, here's one and some people prefer the other
   ... the larger issue i'm concerned with is that every decision that
   we second guess, every little thing that we think have we thought
   this through then it costs us
   ... if this is going to cost an entire telcon then it's not worth it
   ... i am concerned with how long it takes to get to consensus on
   small details
   ... 1) are there going to be knock on effects for html in the future
   (if html introduces a reference mechanism like use)
   ... 2) are there perf issues with this

   ed: the html wg could come back saying we think this has bad effects
   ... there needs to be a discussion on this

   shepazu: we can ask the html wg if these two things will be a

   ed: do we need a decision here today - can we assign the action to
   start the discussion

   <scribe> ACTION: shepazu to send mail to HTML WG raising the issue
   of using SVG in the HTML head [recorded in

   <trackbot> Created ACTION-2924 - Send mail to HTML WG raising the
   issue of using SVG in the HTML head [on Doug Schepers - due

   ed: let's try to get the final things fixed in the test suite
   tomorrow and run the tests next week

   <ed> who's mailing out the minutes today? adrianba?

Summary of Action Items

   [NEW] ACTION: chrisl to bring the 1.2 changes for bidi and text
   anchor back to the 1.1 spec [recorded in
   [NEW] ACTION: chrisl to fix the text-intro-04 test with feedback
   that has been around a little while [recorded in
   [NEW] ACTION: ed to regenerate the test status file and make sure it
   only contains accepted tests [recorded in
   [NEW] ACTION: ed to review the test for ISSUE-2335 [recorded in
   [NEW] ACTION: heycam to add the wording and fix the test for
   ISSUE-2384 [recorded in
   [NEW] ACTION: shepazu to send mail to HTML WG raising the issue of
   using SVG in the HTML head [recorded in

   [End of minutes]

 Chris Lilley   Technical Director, Interaction Domain                 
 W3C Graphics Activity Lead, Fonts Activity Lead
 Co-Chair, W3C Hypertext CG
 Member, CSS, WebFonts, SVG Working Groups

Received on Thursday, 9 December 2010 23:33:38 UTC