[widgets] Draft minutes from 7 October 2010 voice conf

  The draft minutes from the October 7 Widgets voice conference are 
available at the following and copied below:


WG Members - if you have any comments, corrections, etc., please send 
them to the public-webapps mail list before October 21 (the next Widgets 
voice conference); otherwise these minutes will be considered Approved.

-Art Barstow


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

                                - DRAFT -

                        Widgets Voice Conference

07 Oct 2010



    See also: [3]IRC log

       [3] http://www.w3.org/2010/10/07-wam-irc


           Art_Barstow, Robin_Berjon, Steven_Pemberton, Marcos_Caceres




      * [4]Topics
          1. [5]Review and tweak agenda
          2. [6]Announcements
          3. [7]Widget Interface spec
          4. [8]widget: URI scheme
          5. [9]AOB
      * [10]Summary of Action Items

<scribe> ScribeNick: ArtB

<scribe> Scribe: Art

Review and tweak agenda

    AB: a draft agenda was submitted yesterday (
    51.html ). Any change requests?



    AB: October 26 is the deadline for comments re October 5 LCWD of
    Widget Packaging and Configuration (
    [12]http://www.w3.org/TR/2010/WD-widgets-20101005/ )
    ... TPAC: widgets group will not meet; registration after October 22
    results in increased registration fee (
    [13]http://www.w3.org/2010/11/TPAC/#Registration )

      [12] http://www.w3.org/TR/2010/WD-widgets-20101005/
      [13] http://www.w3.org/2010/11/TPAC/#Registration

Widget Interface spec

    AB: Addison Philips submitted a comment against the 7-Sep-2010 LCWD
    28.html ) and it raises the issues about how a developer can
    determine the locale and direction for the "span-able" properties.
    ... Marcos submitted a proposal to address the issues (
    33.html ).
    ... so far, I don't think anyone has responded to MC's proposal


    MC: yes, that's correct

    AB: not clear if directionality needs to be an explicit part of the
    API or if the spec can "punt" on that as suggested by Marcos f.ex.
    by following (
    [16]http://www.iamcal.com/understanding-bidirectional-text/ )

      [16] http://www.iamcal.com/understanding-bidirectional-text/

    MC: I think the API needs some type of extension
    ... eg. the Localizeable DOM String
    ... to add language
    ... thus getters can work

<Marcos> eg. widget.name.lang

    AB: that part seems straight-forward

    MC: the algorithm can be written in JS
    ... don't think we should deal with that at the API level

    AB: so you think the directionality is out of scope for the API
    ... i.e. is handled by some layer above the API
    ... is that correct?

    MC: yes

    AB: anyone else have input on this issue?

    SP: think we need to hear from the I18N WG

    RB: yes, I agree
    ... don't think directionality should be part of the API
    ... for example, it should definitely not be settable

<scribe> ACTION: barstow ask the I18N WG to respond to Marcos'
    proposal for Interface locale and directionality [recorded in

<trackbot> Created ACTION-588 - Ask the I18N WG to respond to
    Marcos' proposal for Interface locale and directionality [on Arthur
    Barstow - due 2010-10-14].

    AB: I'm becoming increasingly concerned the bidi model in P&C is
    overly complicated for a Level 1 spec. Could greatly simplify
    everything by dropping span and just defining the "dir" attribute
    for the span-able elements. Any additional info that is needed can
    be accessed via indirection i.e. include a URI in the metadata.

    MC: in Opera we haven't had any problems implementing it
    ... it is quite simple to process

    AB: the use cases we see for these span-able elements is quite
    ... e.g. just to display the name of a widget in a home screen
    ... or the description is displayed by a widget "store"

    MC: I did a whole lot of research for this

<Marcos> reseach was

      [18] http://dev.w3.org/2006/waf/widgets/i18n.html

    AB: I realize there are use cases for lots of embedded spans with
    different directions
    ... but for a L1 spec we want wide deployement, it seems rather
    ... the other concern I would have here is will we get 2 impls that
    pass all of the span and dir test cases
    ... do the test cases include nested spans with direction changes?

    MC: yes, they do
    ... would it help if I send my JS impl to the list?

    AB: yes, I think it would be helpful
    ... anything else on Interface spec for today?
    ... we'll have to go back to LC to add "lang" support
    ... unless we define the lang in a separate spec

    MC: I think other implementers will provide data about supporting
    P&C bidi model

    AB: that would be good information to have

widget: URI scheme

    AB: what's the status and plan Robin?

    RB: I've been wondering if it should include navigation
    ... and wondering if we coud use or resue the blob: uri scheme that
    is defined in the File API spec
    ... would be cleaner to put navigation in a separate spec
    ... but would be easier to add it to uri scheme
    ... spawning another spec has disadvantages
    ... Need to get feedback from implementors re navigation

    MC: we should do another landscape investigation
    ... some impls don't navigate at all
    ... for instance, Opera doesn't navigate

    AB: you mean intra-widget package is not allowed

    MC: correct, that is not allowed
    ... if click on a link, it starts a new browser

    RB: for V1, wondering if spec should be silent on navigation

    MC: think we can live with that
    ... there are some UCs for navigation

    RB: what if an iframe is included in widget's index.html?

    MC: not sure

    AB: without more data, hard to know if we should say something
    normative, non-normative or remain silent
    ... Marcos, does the landscape doc touch on navigation?

    MC: no, it's a bit dated and didn't look at these more sophisticated
    use cases

    AB: can one of your ask for some feedback?

    MC: yes, I can do that

    AB: ok; great and I'll send in info from the Qt WRT implementation

<scribe> ACTION: marcos Ask implementors for feedback on navigation
    models supported [recorded in

<trackbot> Created ACTION-589 - Ask implementors for feedback on
    navigation models supported [on Marcos Caceres - due 2010-10-14].

    RB: need to know if widget: URI is exposed
    ... how about Opera, Marcos?

    MC: we use it but don't think we expose it

    RB: perhaps I should ask the list if there is value in merging
    widget and blob URIs
    ... but if there are already impls, we don't want to break them
    ... Given the IANA/IETF work involved, would be good to reuse a
    schemem if we can

    AB: it would be good to know if other implementors support widget:
    ... I believe (not certain) that widget URI scheme is implemented by
    the Qt WRT, at least the alpha release
    ... would be good to get implementor feedback on widget: scheme
    ... could use the same email as navigation request

    MC: ok, I'll ask both

    AB: so, registration is on hold pending more implementor feedback.
    Is that correct Robin?

    RB: yes, we need to get more feedback

<Marcos> <a href="iframe.html">navigate to iframe</a>

<Marcos> <iframe width="200" height="200" id="iframe"

<Marcos> </iframe>

<Marcos> <p id="hello"></p>

<Marcos> <script>

<Marcos> var iframe = document.getElementById("iframe").src;

<Marcos> var p = document.getElementById("hello");

<Marcos> p.innerHTML = iframe;

<Marcos> </script>

    MC: when "navigate to iframe" is clicked, the widget does Not
    navigate to iframe.html


    AB: any topics?
    ... next meeting ...
    ... 1 or 2 weeks?
    ... let's go for 2 weeks which is Oct 21
    ... and as always, take discussion to the list
    ... meeting adjourned

Summary of Action Items

    [NEW] ACTION: barstow ask the I18N WG to respond to Marcos' proposal
    for Interface locale and directionality [recorded in
    [NEW] ACTION: marcos Ask implementors for feedback on navigation
    models supported [recorded in

    [End of minutes]

Received on Thursday, 7 October 2010 14:07:28 UTC