[Minutes] MLW-LT call 2013-07-10 and proposal to move forward with "script" topic

Hi all,

minutes of the 10 July call are at
http://www.w3.org/2013/07/10-mlw-lt-minutes.html
and below as text. Also below: a proposal how to move forward with the 
script issue, hence, putting Daniel into CC. All, please reply in this 
thread and say what you think.

Summary: I would propose to close this issue without a change based on 
the following points:

General circumstances & with my co-chair hat on:

1) We had various implementers on the call yesterday and on the mailling 
list saying "we would make the CDATA / XML comments change or don't 
care", see straw poll at
http://www.w3.org/2013/07/10-mlw-lt-minutes.html#item00
"It would be good to see in the minutes votes (just one per particpant) 
like: 1) want CDATA 2) do not want CDATA 3) would be fine with both or 
have no opinion"
But there was no committment about the timeline, see straw poll at
http://www.w3.org/2013/07/10-mlw-lt-minutes.html#item03
"Can we do testing by the end of August? (1) Yes, (2) No, (3) Don't 
know.". Or is the answer of all "don't know"?


2) For the time frame that we are looking at (we should be completely 
finished with the spec including final publication by autumn), this 
situation makes it really hard to implement the change, in terms of 
updating the spec / test suits / implementations. This is also because 
there is no clear consensus about the technical best solution, see 3) 
and 4) below.


Technical reasons (personal view):


3) The discussion that David started within the HTML WG, see
http://lists.w3.org/Archives/Public/public-html/2013Jul/0018.html
http://lists.w3.org/Archives/Public/public-html/2013Jul/0019.html
showed that "we should do the techically best solution" would be ideal 
but it doesn't seem to be possible: both having XML escaped inside CDATA 
/ comments or not escaped (= current behavior) will be difficult for 
tools to handle. The difference is only what tools and tool chains we 
are looking at - processing this in the browser or pure XML workflows. 
See the aspects that are described in the 0019 mail from Robin.


4) A clean technical solution would be to not have an XML representation 
of ITS rules, but rather JSON - again see Robin's mail. That is actually 
something Daniel had mentioned before (more below on that). But that is 
rather something for ITS 2.1 - speaking here again with the timing in mind.


How to move forward? I would propose to document the issue in a note in 
the spec and would ask Daniel to draft the note, if that's ok with you.

Also, I would propose to have a topic in the interest group about "non 
XML syntax for ITS global rules / standoff markup in ITS 2.0". At 
Daniel: in the ITS IG we are discussing topics that are best practices 
or not yet developed for ITS 2.0 - but will give input to ITS 2.x / 3 
etc. You had mentioned before that you would be interested to work on 
such a syntax. Would you still be interested?


I hope that this resolution is acceptable for everybody. Let's see what 
the mailing lists brings and come back to it during next week's call.


Best,

Felix

    [1]W3C

       [1] http://www.w3.org/

                                - DRAFT -

                                MLW-LT WG

10 Jul 2013

    [2]Agenda

       [2] http://lists.w3.org/Archives/Public/public-multilingualweb-lt/2013Jul/0005.html

    See also: [3]IRC log

       [3] http://www.w3.org/2013/07/10-mlw-lt-irc

Attendees

    Present
           Arle, glazou, kfritsche, SebastianS, Ankit, pnietoca,
           Pedro, Milan, tadej, Yves, chriLi, leroy, Des, joerg,
           felix(ircOnly), daveLewis(IRC, only)

    Regrets
    Chair
           Arle

    Scribe
           Arle

Contents

      * [4]Topics
          1. [5]script and ITS 2.0 markup
          2. [6]]init implementation of ITS 2.0
          3. [7]Handling of inline global rules in (X)HTML (again)
          4. [8]topics not handled today
      * [9]Summary of Action Items
      __________________________________________________________

script and ITS 2.0 markup

    <scribe> Scribe: Arle

    Daniel: described the burden of dealing with the script
    solution for ITS rules in HTML/XHTML/etc. Needs switches in the
    code depending on the HTML flavor.
    ... All HTML tools will need to deal with this. Not XML server
    apps. But for HTML it is a big deal.
    ... Outside from ITS, I find it weird to have different DOMs
    depending on whether you use the XML or HTML serializations for
    HTML5.
    ... Solution was to wrap the rules in a CDATA section to ensure
    that we have the same character data nodes and one DOM, no
    matter what the serialization.
    ... We have provisions like this for JavaScript, you have to
    encapsulate the code if you use a < sign.
    ... It's not uncommon.

    Arle: Anyone want to respond.

    Yves: I don't have a strong opinion. I understand Daniel's
    point and I would like to have one representation. But I am
    worried about the change at this point makes so much change.
    LQI and Provenance use the same method for standoff. So we
    would have to do the same.
    ... My implementation wouldn't have to change much. But others
    would. It's late in the process, but we should do the right
    technical thing.

    Daniel: I think if implementations have implemented something
    from a preliminary spec, that is their problem. The goal is to
    make the best possible spec. The DOM difference is a BIG design
    issue, not something minor.

    Yves: Timing is secondary for me. But I'm fine with the change.

    Daniel: Has anyone pinged the HTML WG on this matter to know
    what they know about the DOM issue?

    Yves: I think that Jirka is our main contact there. As he says,
    it is a case of preference to go one way or another, but I
    think Daniel has the technical argument here.

    <scribe> ACTION: Daniel to ping HTML5 WG to ask them about this
    issue. [recorded in
    [10]http://www.w3.org/2013/07/10-mlw-lt-minutes.html#action01]

    <trackbot> Created ACTION-553 - Ping HTML5 WG to ask them about
    this issue. [on Daniel Grasmick - due 2013-07-17].

    NOTE: CHANGE TO DANIEL GLAZMAN, NOT GRASMICK.

    Daniel: If the HTML5 WG doesn't think this is a big deal, I
    will withdraw my comment.

    <pnietoca> I think the same as Yves

    Pedro: I think we need to consider this seriously. I would like
    to stress one unusual factor. Our WG has been funded with the
    mandate to finish before December of this year. The funding
    enabled us to move quickly. In this case, we have show cases,
    final clients, etc.
    ... My only question is, if we can deal with this without
    putting the commitments of the project at risk, how will this
    affect things. If we can fix this in 2.1 perhaps?

    Daniel: Basically, you would not change the technical contents
    of the spec, but to fix one minor issue. It would mean fixing
    five to ten tests that could be easily done. I imagine you have
    XHTML and HTML tests separately. So you could remove some tests
    and consolidate.
    ... Who is the activity leader? I'll ask Richard what he thinks
    of it.
    ... It might add a few weeks in the loop, but it shouldn't blow
    your schedule.

    Pedro: For us it is important to make sure we match the EU
    timing requirements.

    Daniel: I'm hitting the issue now because of the EU funding to
    work on this.
    ... It is a question of complexity, code maintainability,
    saving an HTML doc into XHML (and back), changing flavors, etc.

    <fsasaki> hi all, for the record, please don't decide about the
    change today. But have a straw poll. It would be good to see in
    the minutes votes (just one per particpant) like: 1) want CDATA
    2) do not want CDATA 3) would be fine with both or have no
    opinion

    Yves: Could we hear from Karl or Pablo as implementers? Do they
    see it as a big issue?

    Pablo: For me it is not a big issue, but I think it would be
    better to save trouble with CDATA, but it's not big.

    Karl: I didn't try it yet, so I don't know for certain, but I
    don't imagine it is big.

    Daniel: There is a fourth possibility beyond Felix's 3: I ping
    the HTML5 WG and ask them for input.

    <pnietoca> I agree with Daniel on this last one

    <joerg> Pinging the HTML5 WG should definitely be the first
    action

    <daveL> for TCD we would vote 3) would be fine with both or
    have no opinion

    Des: I'm trying to catch up with this. Just to be clear,
    wrapping the rules in CDATA, there is no localizable content
    wrapped in CDATA? Correct?

    Daniel: Correct.

    <Ankit> for DCU: 3)

    Daniel: That way, you just need the first child of the <script>
    element and you get it and then you have one way of
    manipulating it.
    ... In fact if we could have the ITS rules subtree in HTML, I
    would prefer that, but we can't. They aren't parsed. We can't
    get them.

    Arle: Poll on the IRC.

    Option 1: want CDATA.

    <glazou> glazou, 1

    Option 2: Don't want it

    <Yves_> for ENLASO: 3

    <tadej> 3

    <leroy> TCD 3

    Option 3: Fine either way or no opinion.

    <kfritsche> 3

    <SebastianS> 3 for ]init[

    <pnietoca> I vote 3) both options are fine with me

    <Des> 3

    <fsasaki> hi all, again,

    <Milan> Milan 3

    Arle: Seems there is no consensus. HTML WG would offer value
    here.

    <fsasaki> just to make sure: are people aware that 3) will
    include people need to update tests and implementations? would
    be good to know that too, for the IRC record(.

    Felix: That was discussed before you appeared on the IRC.
    People seem to be aware.

    <pnietoca> I don't mind updating the tests

    <fsasaki> thanks arle, could people say again: 1) ok with
    updating my tests 2) not ok with updating tests. thanks

    <glazou> thanks people

    Arle: Another poll, per Felix's request.

    <Yves_> Enlaso: 1 (ok with updating the test)

    <pnietoca> 1) ok with updating my tests

    <tadej> 1

    <leroy> 1

    <Ankit> 1) ok with updating tests

    <Pedro> 1

    <Des> 1 (We're not fully implemented anyway

    <Milan> 1

    <Pedro> Des, very smart!

    <Des> Lol

    <kfritsche> 1 okay with testing (but not sure about in which
    time frame) [sorry for late post]

]init[ implementation of ITS 2.0

    Sebastian: I have forwarded our import and export files to the
    group. Would ask for input. We have a freeze until Monday.
    Thanks to Yves for helping us spot one problem.
    ... From the beginning, we use translate/no-translate (to match
    mtype="protect")
    ... We use annotations (scribe: some things missed here)
    ... We didn't find in the XLIFF mapping what mtype to use for
    locale filter. We also used comments for some other things.
    ... What namespace to use for localization note.
    ... We use XSLT transformation and XParser for Java. We have a
    theoretically unlimited way of nesting elements (limited only
    by memory). For XLIFF we use Okapi.

    Sebastian: I've sent the ODT and XLIff file. There are also
    binary files in our deliverable.
    ... Information missing is in extra files so we have a valid
    XLIFF file conforming to the schema.

    Yves: A few other notes. I'm looking at the ODT_XLIFF file
    ... You have a namespace called itsxliff that matches the its
    namespace definition. I think it should match itsx.

    Sebastian: We use Okapi but had to tweak it a bit. Extended the
    filter for XLIFF Writer.

    Yves: Just so you know, on the xliff element the namespace
    seems wrong.

    <Yves_> <mrk mtype="x-its">vitae porta</mrk>

    <fsasaki> hi all again. Arle: can you come back after the
    ]init[ presentation and ask again whether the people are ok
    with changing testing by end of August? The timing is critical.
    And getting a clear "yes or no, I can do the testing in the
    given time" is essential

    Sebastian: We had to use that because the mapping is marked as
    "to do".
    ... If you have time/motivation, send us your comments.
    ... We can make simple changes very quickly.

    Christian: I have a completely different issue. No looking at
    the document, I wonder if/when we might be able to provide
    additional details.
    ... Would your work become part of the official Libre Office
    distribution, or do you have a different mechanism in mind?

    Sebastian: We will use LGPL Libre Office extension.
    ... We have to discuss bundling?
    ... If nobody has a problem with LGPL, we will link to Okapi,
    but don't want to replicate Okapi.

    Christian: What happens if Okapi progresses but your extension
    does not? Will it mean the extension will be stuck in an
    outdated version of Okapi?

    Sebastian: two possibilities: (1) separate downloads; (2)
    bundled download in one place.
    ... We have not only ITS ODT, but XLIFF reusing existing work.

    Christian: Also, note that we want to create general outreach
    videos. Having some related to the tools would be nice. If you
    already have one or could make one, it would be cool.

    Sebastian: Yes. We have a video unit.

Handling of inline global rules in (X)HTML (again)

    Poll question: Can we do testing by the end of August? (1) Yes,
    (2) No, (3) Don't know.

    Des: It depends on when the spec is updated.

    Felix, anything to add here?

    <fsasaki> hi all again, people should be aware that this change
    means: delaying ITS 2.0 proposed recommendation until mid
    august and final recommendation until end of september - at
    least

    <fsasaki> so timing is critical for doing this. and here we
    cannot forget the EU funding which ends end of December.

    Pedro mentioned this issue, Felix.

    <fsasaki> thanks, Arle

    Arle: I think we can't solve this in five minutes.

    Karl: Let's take it in email.

    <SebastianS> May I ask you, whether there is a deadline for the
    XLIFF-Mapping (esp. for LocaleFilter and LocalisationNote)?

    Yves: We are supposed to have the deliverable at the end of
    September and there is a lot of mapping to be done, but I hope
    to have it by the end of August.

    Sebastian: I'm asking because our internal deadline is July, so
    we may have to implement a few categories with placeholders in
    the meantime.

    Yves: We may be able to prioritize the categories you need. We
    will get something stable for locale filter and localization
    note.

    <fsasaki> hi all again, will you still answer the poll question
    during this call? "Can we do testing by the end of August? (1)
    Yes, (2) No, (3) Don't know.". Or is the answer of all "don't
    know"?

    Sebastian: Terminology is an issue. Translate will use
    protected.

    It seems the consensus is "don't know"

    <Pedro> The main iussue about the CDATA change is if we will
    reach a moment where we have to decide: changing and not
    getting the EU commintment, or getting the EU commintment and
    changing later. I hope both things are compatible.

    <fsasaki> thanks, Arle - that's an interesting "consensus" :)
    but good to know - it helps to estimate the realism of really
    doing this change. I will write a mail to the list later.

    <fsasaki> Pedro, what do you mean by "EU commintment"?

    We want to get feedback on this by next week. Not sure who to
    assign the action to for this.

    <glazou> Pedro, again, I think we can live with the current
    spec if that's a too expensive change, but you guys have to
    understand and probably note in the spec it's a burden on
    implementors...

    <fsasaki> you can give the action to me

    Felix, we want to resolve this quickly, but there were too many
    uncertains.

    <scribe> ACTION: Felix to collect feedback on whether testing
    can be completed by the end of August. Due July 17. [recorded
    in
    [11]http://www.w3.org/2013/07/10-mlw-lt-minutes.html#action02]

    <trackbot> Created ACTION-554 - Collect feedback on whether
    testing can be completed by the end of August. Due July 17. [on
    Felix Sasaki - due 2013-07-17].

    <Pedro> I mean to have as outreach of the LT-Web project an
    stable and approved ITS 2.0 that allow to disseminate and
    impementing in real life.

    <fsasaki> thanks for noting that, glazou, very helpful. note to
    all: I will also check to see what people in the IA domain
    think

    <Yves_> Just a reminder: the change would affect also LQI and
    Prov standoff.

    <fsasaki> Pedro, understand, Yves_ thanks for the reminder

    <fsasaki> close action-553

    <trackbot> Closed ACTION-553 Ping HTML5 WG to ask them about
    this issue..

    <Pedro> Thank you, glazou.

    <glazou> np

topics not handled today

    Tabling XLIFF mapping and other ITS IG WG work items for next
    call.

    Also, for the agenda next time, LQI and MQM.

    <fsasaki> unfortunately regrets for next week - I have to be in
    disneyland (tokio) ...

Summary of Action Items

    [NEW] ACTION: Daniel to ping HTML5 WG to ask them about this
    issue. [recorded in
    [12]http://www.w3.org/2013/07/10-mlw-lt-minutes.html#action01]
    [NEW] ACTION: Felix to collect feedback on whether testing can
    be completed by the end of August. Due July 17. [recorded in
    [13]http://www.w3.org/2013/07/10-mlw-lt-minutes.html#action02]

    [End of minutes]
      __________________________________________________________


     Minutes formatted by David Booth's [14]scribe.perl version
     1.138 ([15]CVS log)
     $Date: 2013-07-11 05:41:17 $

      [14] http://dev.w3.org/cvsweb/~checkout~/2002/scribe/scribedoc.htm
      [15] http://dev.w3.org/cvsweb/2002/scribe/