[minutes] Internationalization TPAC breakout session 2017-11-08

https://www.w3.org/2017/11/08-i18n-minutes.html




text version follows:


                                - DRAFT -

                              TPAC breakout

08 Nov 2017

Attendees

    Present
    Regrets
    Chair
           SV_MEETING_CHAIR

    Scribe
           chaals2

Contents

      * [2]Topics
          1. [3]intro
      * [4]Summary of Action Items
      * [5]Summary of Resolutions
      __________________________________________________________

intro

    <chaals-o> r12a: Beep

    <chaals-o> r12a: The i18n mission is to make the web work
    worldwide. That's a big job and there are other groups doing
    bits of it.

    UNKNOWN_SPEAKER: We look at things like CSS, Timed Text, HTML,
    ... and they need to be able to represent text in a way that is
    recognised as natural by the local users

    <scribe> scribenick: chaals2

    UNKNOWN_SPEAKER: Japanese has wakiten, which the west does not.
    Arabic stretches text rather than spaces for justification.
    ... these are not easy today - and there are complex rules
    about what you can stretch anyway.
    ... We need to understand the requirements, and then make sure
    W3C work can deal with it.
    ... And it is hard to find english-language explanations of
    requirements for languages that are not english, like Thai or
    Sinhalese
    ... Once upon a time we didn't make a big effort to look for
    experts around the world, beyond the ones we had. That changed
    starting with the Japanese Layout Task Force.
    ... They worked and wrote in Japanese, and then translated
    their output into english
    ... describes the unique (or different from english layout?)
    requirements for layout.
    ... So how do you support different appraoch to layout in CSS,
    etc.

    [r12a describes ruby usage and layout as a detailed example]

    r12a: We get the layout requirements, and then we need to find
    out what implementations do and what is not supported.
    ... E.g. people say "yes we can do vertical layout" - but then
    it turns out that there are unmet requirements, like being able
    to have latin text included in several orientations at once.
    ... Now we have a number of requirements documents, and there
    are more that are in development - and probably more that we
    should get.
    ... It is hard. People get enthusiastic and start, and then
    realise that they need to do a lot of work (write a book,
    roughly) to describe their requirements, and they slow down.
    ... We are thinking the best way is to start with the gap
    analysis - what doesn't work today?
    ... and then prioritise the gaps, and write that bit up, and
    some tests, and focus on getting implementors to do that. And
    proceed iteratively

    <Ralph> Chaals: another approach is to go to a university and
    recruit a student as a PhD project??

    r12a: I would prefer to have a group of people than a single
    student. E.g. Arabic has multiple possible approaches to
    justification, and one single person may not cover that well.
    ... Ethiopic lreq is done by one person, and we are concerned
    about lack of review. But it is something to think about.
    ... Right now we are trying to work out how many gaps there
    are.
    ... we have a matrix
    [6]https://w3c.github.io/typography/gap-analysis/language-matri
    x.html
    ... this is currently a quick draft, not validated.
    ... There are things that people expect for everyday use,
    things that are not needed on a daily basis for writing, but
    important for e.g. basic publishing quality.

       [6] 
https://w3c.github.io/typography/gap-analysis/language-matrix.html

    jacques Are the question marks things we don't know?

    r12a: Right.
    ... And there are things marked as "it does not work" because
    it doesn't.
    ... then we have more detailed explanation for each language

    [shows amharic as an example]

    r12a: We also have a docuiment on layout and typography - CSS
    doesn't want to define every single thing about how to handle
    layout, they want implementors to be able to do it, but this
    provides implementors with a reference to the requirements we
    know of.
    ... There are known unknowns in there, and we have listed them
    as issues in a github repo

    [r12a describes the index of pictures of things that happen]

    CMN: Have you considered stuff like braille layout, sign
    languages, etc?

    r12a: Not yet, but we can put it in there.

    jacques Is there a test suite to help people detect what is
    wrong?

    scribe: e.g. we have automatic translation. Can we get some
    text that shows up the problem when you put it in a translator?

    r12a: There is the Universal Declaration of Human Rights in 400
    languages - but it is a specific format. But basically, no we
    don't have that. We would like people to come to us and say
    they have experienced some problem.

    @@2: Do you have indicators that CSS will cause problems in
    specific languages - a checklist that you can say what is going
    to go wrong?

    r12a: Sort of. We have some tests that can sort of be used for
    that...

    @@3: Work done for CSS has been based on requests?

    r12a: Yes, or - currently mostly - people who happened to know
    something was needed.
    ... We lack a way to find people who are experts in publishing
    and layout, and know what people want to do

    jacques You don't want people to write about CSS...

    r12a: Yes. We want to describe the requirements, not the
    current state of the art.

    jacques Want to avoid trying to find an expert in CSS and some
    language

    r12a: We had someone wanting to do gap analysis in the
    requirements document - or accepting what they think is "good
    enough" and not including the actual requirements.

    @@4: regarding security. There is a discussion about homoglyph
    phishing. Does this group have scope to look into that issue?

    r12a: Other people are working on it and we don't have
    capacity, so we don't work on that.

    @@5: how do you choose which language to work on?

    scribe: Mongolian is not used in everyday settings.

    r12a: Right. 2.2 million people speak mongolian in Mongolia,
    generally use cyrillic, but would like to use their own script.
    ... the matrix has the top 100 languages from wikipedia by
    usage.
    ... Which is not really right. There are some unique things for
    Mongolian.
    ... In my mind it is important to get it into the list because
    it has architectural properties that we will not solve if we
    don't include it in our set of requirements. Same for Javanese.
    ... So there is a mix of what people want to pay for, whether
    something is endangered, whether there are unique technical
    requirements implied.
    ... And where we can find someone who can tell us something.

    jacques 2 major reasons. Want people who speak a language to be
    able to use the web, and want to use their language, but also
    to make sure a language doesn't disappear as people just shift
    to english.

    r12a: We are going to try and help people where they come with
    requirements. We can make contacts in implementation, technical
    development, etc. But we need subject matter experts.
    ... And it is hard to convince people that their ordinary
    knowledge is actually information they have that the rest of
    the world doesn't.
    ... So they don't feel confident that they should contribute.

    Jacques: How do i18n pages switch language? What's the magic?

    r12a: Usually someone has manually done the work - or checked
    it.

Summary of Action Items

Summary of Resolutions

    [End of minutes]

Received on Wednesday, 15 November 2017 15:53:17 UTC