Internationalization Working Group Teleconference -- 20 Oct 2020

<addison> trackbot, prepare teleconference

<agendabot> addison, sorry, I did not recognize any agenda in https://www.w3.org/wiki/I18N_2020_TPAC

<addison> Additional Agenda Info: https://lists.w3.org/Archives/Member/member-i18n-core/2020Oct/0031.html

<addison> 1400 UTC

<atsushi> will switch shortly (also from AC)...

<r12a> from Alan at CSS:

<r12a> @r12a would you be available for any of our remaining meetings this week to discuss this issue?

<r12a> https://wiki.csswg.org/planning/tpac-2020#schedule

<r12a> https://github.com/w3c/csswg-drafts/issues/3029#issuecomment-712471470

Agenda and Getting Organized

<addison> welcoming Vainateya from CDAC

<addison> vainateya: interested in how to take Indic forward

<addison> ... have researchers and multinationals

<addison> ... onboard and trying to figure out pre-standardization over there

<addison> ... get in touch with richard and other experts

<addison> ... looking at IIP

<atsushi> (let me just listen for IIP...)

<addison> indic script topics

<addison> internationalization considerations (maybe thurs?)

Indic Script Work

<Abhijit> Yes

<scribe> scribe: xfq_

Abhijit: very briefly, when I worked in IBM
... while I was in IBM for me @@ as main language
... for me W3C is important
... W3C develops open standards
... when you talk about users not understanding english at all
... not political point, about execution
... 4 years ago, discussions about what could be monetized
... 5,6 billion dollars operation

addison: what we might be able to help you with?
... lots of interests in different scripts, local languages in India
... jlreq is very successful
... interested to develop layout requirements for Indian?

Abhijit: from an engineering perspective
... ISCII like ASCII
... we want unicode to be adopted
... encoding is one layer
... I/O is another layer
... the specs are not bad
... we can operate Indian languages seamlessly
... but they're not implemented in standard ways
... not everybody is digitally literate
... according to me @@ is not designed for transactions
... indivival characters can be constructed in different ways
... normalization not impl'd correctly

r12a: same problem with Thai, Arabic, Khmer
... combining characters

addison: not recomposed
... seem to be the right shape on the screen
... order of code points are different

Abhijit: @@

r12a: normalization does not solve these issues

addison: we don't see this problem in Western European languages in the way that we do in other languages
... some ambiguity about the sequence

Abhijit: banking in these languages
... disadvantage

addison: not necessarily evil
... diversity of encoding is not helpful

Abhijit: for identity papers
... for people cannot read english
... or for election document

addison: documenting those issues is one thing
... where you would go to address the different kinds of issues
... probably some stuff the Unicode Consortium can do
... probably some stuff system vendors can do
... the challenge is whether it's in our scope
... i think documenting the gaps is in our scope

Abhijit: @@

addison: I think vendors would be incentive to do this for customers
... i'm interested in which languages are of most interest

Abhijit: touchy topic in India
... come up with clear issue that needs fixing and decide how you want to fix the issue

r12a: very little is happening in terms of "let's write something down, 1, 2, 3, 4"

Abhijit: Hindi is a language
... Assamese Bengali
... it's genuine requirements
... I'm not a linguist
... the problem is well-known

r12a: I would like to help you
... I could probably write some of this down
... not enough bandwidth
... I would be happy to review
... how do we find experts and how do we engage with them?
... normalization is one issue
... language matrix is really what all of the issues that are preventing people using the web
... we have the classification system

Abhijit: NYT article about WeChat
... one platform in China

addison: I think our challenge is that we're not necessarily Languages experts either

<Abhijit> I'll mark you on the link.... of NYT

addison: we need help from people who speak the languages and understand the issues

Abhijit: the plan is concrete in my opinion

addison: I do think that the major platform owners (Google, Apple, Microsoft etc.) are very interested
... for their customers
... if we can get the information to them if you make progress

<r12a> Vainu, have you seen this document? https://w3c.github.io/i18n-drafts/getting-started/languagedev.en If you follow the links from it, it should give an overview of the Language Enablement Framework.

Abhijit: "this is our platform. this is a particular error in this language. this is another error. etc."
... our goal is not to make standards
... but to show what is required

addison: I personally have not been on the front of organizing this
... i know r12a spent a lot of effort over the years trying to organize, develop, deliver the materials

<Abhijit> This is the post on LinkedIn.....

<Abhijit> https://www.linkedin.com/posts/deepblue_how-china-is-changing-your-internet-the-activity-6710838052832473088-DyUa

vainu: based on data
... @@ static html page in 22 languages
... basically we looked at many languages, on government service or book a ticket
... in India, the Indian language implementation is not uniform
... transaction info may be in English and the menu is in the local language
... local language is used in these cases

Abhijit: at the end of the day it is an engineering problem
... if somebody had taken the trouble to engineered these things properly

addison: I'll take jlreq as an example
... the Japanese layout task force spent a lot of time documenting how Japanese text layout works, all the details of it
... Japanese mostly works, except a few things
... a challenge in these conversations is that there's a tendency to be overstated
... in most cases you can get it work
... the problem is not there's no support for the language
... the problem is some gaps
... need actionable items when we talk to CSS

Abhijit: agreed
... what i'm trying to convince people is the benefits of doing it right
... that's pretty much all I was going to say

addison: do we have more things that we would like to do in this space?

r12a: we can't fix problems @@

CSS Issue discussion

<r12a> https://wiki.csswg.org/planning/tpac-2020#schedule

r12a: I thought it might be fit to go over the CSS issues

<r12a> https://github.com/w3c/csswg-drafts/issues/3029#issuecomment-712471470

r12a: ^ issue
... for some arabic text you declare the left hand margin to be 10px

<r12a> https://w3c.github.io/i18n-tests/css-logical/inheritance/inherit-logical-logical.html

r12a: ^ here's a test
... if a RTL div contains a LTR div
... that has margin-inline-start set to inherit

[r12a introduces the issue and the test]

addison: your assertion is because it inherits
... @@

r12a: that's the question

addison: I don't see a lot of argument on that thread

r12a: this is the issue that is holding up a lot of logical properties and stuff that we talked about before

addison: do you need additional people from CSS to discuss this?

r12a: yes

addison: argument could be made for either of those inheritance models
... doesn't make a lot of sense to me
... interpreting the rule locally rather than inheriting the interpreting rule

r12a: purely logical approach that one could take
... if I said here's a margin for Arabic and it's the start margin, and then I say that i want to inherit the margin
... interoperably do it the other way so far

fantasai: what happen if css had been designed as logical from the start
... rather than physical as the start

myles: i think the example that r12a just gave doesn't behave that way in any browser

r12a: that's correct

<Abhijit> If there is anything you would like me to type, please let me know.

r12a: I have just thinking it through again
... I would get confused it that's not the case because I'd have to do that conversion

addison: wouldn't know who started it without tracing it back to the specific inheritance level

<Abhijit> ( Sure. Please let me know if you need me to expand on something later. I would be happy to do that. )

myles: agree with r12a but given the fact all the browsers have interop
... bunch of tricky questions that would have to be answered
... 3 ways
... @@
... 2. logical everything
... 3. carve out some small situation

fantasai: 4. inherit depending on property
... no web compat issue
... the child says "i want to inherit"
... look at margin-inline-start
... don't know if it's easy to implement in the styling system of the browser

myles: I think that makes sense
... decisions need to be intentionally made

fantasai: determined by the cascade
... if i say "margin-inline-start: inherit" I would expect that to be have inherited
... @@
... before you cascade them according to the writing mode, @@

addison: if the logical was the last one does that inherit down as logical?

fantasai: no

addison: just inherit the rule down

fantasai: that's what I was suggesting
... if you two have the same computed value it doesn't matter who set it they're required to have the same computed value in the end
... @@

addison: OK

r12a: so have we solved the problem?

fantasai: question is @@
... there's no asymmetry in how logical and physical properties are handled in this proposal, but there clearly are in implementations

addison: do we need to write down some details?

myles: I think the last part of fantasai's comment should be written down

fantasai: I feel like diagrams would help

addison: do you need action items?

<addison> ACTION: fantasai: document dicussion of logical vs. physical inheritance from I18N TPAC meeting

fantasai: I can do it

<trackbot> Created ACTION-968 - Document dicussion of logical vs. physical inheritance from i18n tpac meeting [on Elika Etemad - due 2020-10-27].

<fantasai> https://wiki.csswg.org/planning/tpac-2020

[Discuss remaining CSS meeting time slots for discussing CSS-i18n issues]

addison: Thursday, fallback to Friday, will ping CSS by email

r12a: fantasai is going to be crucial to our discussion
... also Myles

fantasai, myles: we'll be at all CSS meetings

addison: anything else on this topic?
... any additional issues to discuss with CSS today?
... fantasai, myles?

fantasai: let me see the issue list

search 'css' in https://www.w3.org/PM/horizontal/

<fantasai> https://github.com/w3c/csswg-drafts/issues/4606

https://github.com/w3c/csswg-drafts/issues/4910

fantasai: re logical property issue, @@ should be encouraged by default

<fantasai> https://twitter.com/fantasai/status/1318578343105556482

fantasai: red parti is what we're going to do
... the blue part is what browsers are currently doing

r12a: followed a few talks in the Unicode conference
... one was good, one was confusing to me
... touched on topics that we've been going through today

<addison> ACTION: addison: write to iuc organizers about making presentations public

<trackbot> Created ACTION-969 - Write to iuc organizers about making presentations public [on Addison Phillips - due 2020-10-27].

addison: generic font issue?

r12a: interested in what the jlreq folks are doing about character classes

atsushi: we had a group meeting in Japanese today
... reorganize character classes into two groups
... one is for describing general layout method of individual characters
... the other is for layout with specific layout method, could differ from layout of their original character class
... character classes was from JIS

r12a: what would the end result be in terms of the document?

atsushi: consensus is that we don't want to edit the current document @@
... we need to carefully go through the documents

addison: is the problem to do with a mismatch between the JIS and Unicode character class models?

atsushi: yes
... in some cases like Western characters in JIS there are two styles of characters, ASCII and fullwidth characters
... @@ writing mode
... complicated mapping between JIS and Unicode

addison: I saw Eric Muller's long analysis about line joining (insertion of or non-insertion of spaces when joining lines)
... any other issue?

atsushi: @@

<fantasai> https://www.w3.org/TR/css-inline-3/#valdef-initial-letter-align-leading

fantasai: initial-letters

<fantasai> https://www.w3.org/TR/ilreq/#h_scripts_without_hanging_baseline

fantasai: for languages that have a bar on the top
... want to see if that interpretation is correct

Abhijit: it's well understood what this means

r12a: would be nice if Elika raise an issue in the iip issue list

https://github.com/w3c/iip

scribe: we should remove that diagram in ilreq

fantasai: what's the problem with ilreq?

r12a: quite a few issues
... one of the main things is segmentationos

Abhijit: @@ markup, usability of the end result

fantasai: question is about where does the first letter start and end
... in english drop cap
... as an occatiooon a cap height @@
... if you increase the amount of spacing between the lines of text
... @@
... then we need to revise the spec

<r12a> Raise issue here, Fantasai, https://github.com/w3c/iip/issues

Abhijit: most native users would know how to interpret it because it is decorative

fantasai: browser has to be responsible for the size
... ink top/bottom in english is cap height and the alphabetic baseline

Abhijit: my opinion is not reliable but I don't think it matters
... Punjabi for example has three lines, Bengali has two lines

fantasai: the diagram in 5.2 is very different from the one in 5.4

Abhijit: point of the expansion @@

r12a: if you look at the pitcure in 5.4

[r12a shares his screen]

r12a: the top of the paragraph runs through the top of the large character
... what would happen if you @@
... would the whole thing shrink down
... or would it remain pretty much as it is now

[Vainateya shares his screen]

r12a: there's a class of script that has the line, but some scripts do noot

fantasai: generally typographers would like to have consistency for the font size
... in the Odia example @@

[fantasai shares her screen]

fantasai: the designer want to align the op/bottom the letter
... you can choose the number of the lines
... in order to size the letter correctly
... we need to know the landing point at the top/bottom

Abhijit: it's purely decorative

r12a: going back to the Bengali example
... this is an example of a script with the head line
... looked at a lot of examples of these things

fantasai: even in en/fr if there are accent in capital letters
... they're ignored

Abhijit: a printed character can be implemented in any way they want
... in a font @@
... bottom is cut off

r12a: problem is whether this is a realistic example
... or just badly put together

Abhijit: looks like a mockup

r12a: thanks for your participation every body!

- DRAFT -

Internationalization Working Group Teleconference

20 Oct 2020

Attendees

Contents

Agenda and Getting Organized

Indic Script Work

CSS Issue discussion

Summary of Action Items

Summary of Resolutions

Scribe.perl diagnostic output