W3C

– DRAFT –
Internationalization Working Group Teleconference

22 July 2021

Attendees

Present
addison, Bert, Richard
Regrets
Atsushi, Felix, Fuqiao
Chair
Addison Phillips
Scribe
Bert

Meeting minutes

<addison> trackbot, prepare teleconference

Agenda Review

Action Items

<addison> https://www.w3.org/International/track/actions/open

r12a: Not done action-1039 yet.

<addison> https://github.com/w3c/string-meta/pull/56

addison: Did not yet do 1050 yet, but will soon.
… could r12a review what I did direction?

r12a: Call it "paragraph direction"? And then I will add that term and definition to the glossary.

addison: The Glossary should contain definitions and then point to other docs.

r12a: Only if there is additional info in that other doc.

addison: The Glossary could indeed orignate terminology.

r12a: That will be the case with "paragraph direction".

r12a: I've been talking with Marcos about linking to glossary items.
… He has a kind of references system.
… Do we put things in that database? It is a syntax with square brackets.
… There could be a database of our own, using a variant of that syntax.
… I will investigate.

Action: richard: investigate glossary linking solutions, e.g. Marcos's

<trackbot> Created ACTION-1052 - Investigate glossary linking solutions, e.g. marcos's [on Richard Ishida - due 2021-07-29].

addison: Did not do action-1051 yet.

Info Share

No info to share.

RADAR Review

<addison> https://github.com/w3c/i18n-request/projects/1

addison: We got a request to review DOM. Who wants to do it?

Fuqiao did the previous DOM review.

addison: I will assign Fuqiao for now.

"Awaiting Comment Resolution" reviews

Charmod-norm

<r12a> https://w3c.github.io/bp-i18n-specdev/#string_match

r12a: I went through the issues mentioned on the agenda and thought about improving specdev.
… There are matching sections, but no matching dos and don'ts. Should we add those?

addison: There is mustard in charmod. Does that not appear here?

r12a: Not sure if the latest changes in charmod have been taken into account in specdev.

<addison> http://aphillips.github.io/charmod-norm/#charmod_n11n_should_not

addison: charmod in many case more strongly worded.

specdev section 5.3 seems mostly verbatim from charmod, but specdev section 5.2 is not.

r12a: I made some content for specdev, should we interleave that with text from charmod-norm?

addison: charmod-norm has algorithm and choices for spec writers. specdev's text is not wrong, but it should diretc people to go and read charmod-norm.
… With the text in specdev, people will not necessarily arrive at the right matching algorithm.

r12a: That is the case for all mustard in specdev. There are always links to get more information.
… specdev is a succinct story. Our discussion has always been how much should be in it.

addison: I want our best practices documents to have the correct best practices and have that reflected in specdev.
… My concern here is that we have some best practices that are kind of high level. That should be explained.

r12a: I think I'm saying the same thing, but I have a different solution. Let's look at sections that follow. Specdev has mustard, but here is nothing in charmod-norm.
… So we can copy it to charmod-norm and then the two can be the same verbatim.
… In other words: I think there is content missing in charmod-norm.

addison: That could be true.

<addison> http://aphillips.github.io/charmod-norm/#CompatibilityFoldNormalizationStep

addison: There is a big red box. Maybe we meant to have mustard there.
… Do you want to make a pull request?

r12a: specdev has had the mustard for some time, but the links were not very useful. I found linking to the right place was actually a bit tricky.
… There is a bit of information at the end of the link and then links to other sections.
… It is easier for specdev to point to one location that has everything you need.
… charmod-norm section 3.2.1 seems like a short section, but actually there is more, and then more after that...

Action: richard: make PR for charmod-norm with proposed additions to mustard and organizational updates to section 3.2

<trackbot> Created ACTION-1053 - Make pr for charmod-norm with proposed additions to mustard and organizational updates to section 3.2 [on Richard Ishida - due 2021-07-29].

r12a: We also should think about publishing, esp. charmod-norm.

addison: Let's try getting through this round of editing. If it takes long, we can publish a snapshot.

r12a: There is still an outstanding pull request.

<r12a> https://github.com/w3c/charmod-norm/pull/215/files

addison: Any reason not to merge it?

r12a: We can make the patch perfect and then merge, or merge and then improve the doc. I think the latter is better here.

addison: OK, I will merge it now.

<r12a> you have:

<r12a> <span class=codepoint>&#x1F88;&#x0301;</span> [<span class=uname>U+1F88 GREEK CAPITAL LETTER ALPHA WITH PSILI AND PROSGEGRAMMENI</span> + <span class=uname>U+0301 COMBINING ACUTE ACCENT</span>]

<r12a> but it should be

<r12a> <span class="codepoint" translate="no"><span lang="el">&#x1F8C;</span> [<a href="/scripts/greek/block#char1F8C"><span class="uname">U+1F8C GREEK CAPITAL LETTER ALPHA WITH PSILI AND OXIA AND PROSGEGRAMMENI</span></a>]</span>

r12a: Missing lang attribute, missing do-not-translate attribute and the codepoint tag should be around the whole thing
… Not yet fixed in the document.
… I used character references instead of characters, because the characters got normalized by the editor program.

addison: I normally use character entities precisely for that reason.
… Do you fix the examples or should I? Can you make a pull request?

Awaiting comment resolution

<addison> https://lists.w3.org/Archives/Member/member-i18n-core/2021Jul/0014.html

r12a: I have another issue on charmod-norm:
… There are situations, e.g., in Kasmiri, where half the people use precomposed letters and half use non-spacing sequences, and I find nothing in Unicode to normalize them.
… Unicode says the precomposed character is only used in Kashmiri. Should we say what to do in that case?

addison: I think Unicode should.

r12a: Unicode model is based on semantics, but we see people use them graphically: use characters that look identical but that Unicode says not to use for that language.
… E.g., a guy in India, knowledgable in principle, still often asks me how to do something in Kashmiri. There is total confusion.
… Some things are orthography specific. Cataloguing everything is difficult. But somehow people have to write Kashmiri.
… Font not always supports it.
… E.g., there is an arabic diacritic that looks just right for Kashmiri, but it is not meant for Kashmiri.
… Thus searching and normalizing in Kashmiri only half works.

addison: Asking Unicode may lead to fallout. Kind of corner cases. If you don't write Kashmiri, you'll never notice. But if you write Kashmiri, you encounter it all the time.

r12a: It is not just Kashmiri. Also happens in Bengali. You can use the wrong characters but that look right.
… I think charmod-norm should at least have a note to say that even after following the ruls in the spec, there are still areas with problems.

addison: We can add text. This seems to be mostly targeted at document authors, because there is no algorithm and thus authors have to take care themselves.
… There are different keyboard implementations, also for Vietnamese, e.g., and people don't necessarily see what they type.

<addison> https://lists.w3.org/Archives/Member/member-i18n-core/2021Jul/0014.html

addison: There are four items here ^^ without issues and that I think can be closed.

<addison> https://github.com/w3c/i18n-request/projects/1

addison: I used are the column "awaiting resolution" on the radar, from bottom to top, to see which items can be closed and moved to "completed".

r12a: The UTR#53 issues are stale. Let's close.
… For tracking-dnt, not sure. Maybe close and recycle?

bert: No WG for it at the moment, so probably recycle when there is a WG again.

r12a: Who is closing the issues?

addison: I am.
… I'll close the other non-contrversial ones as well.

Summary of action items

  1. richard: investigate glossary linking solutions, e.g. Marcos's
  2. richard: make PR for charmod-norm with proposed additions to mustard and organizational updates to section 3.2
Minutes manually created (not a transcript), formatted by scribe.perl version 136 (Thu May 27 13:50:24 2021 UTC).

Diagnostics

No scribenick or scribe found. Guessed: Bert

Maybe present: r12a