Re: [csswg-drafts] Font-family name matching requires full Unicode case comparison (#3675)

The CSS Working Group just discussed `Font-family name matching requires full Unicode case comparison  (reconsider FTF resolution)`, and agreed to the following:

* `RESOLVED: Void the previous resolution and close no change`

<details><summary>The full IRC log of that discussion</summary>
&lt;dael> Topic: Font-family name matching requires full Unicode case comparison  (reconsider FTF resolution)<br>
&lt;dael> github: https://github.com/w3c/csswg-drafts/issues/3675#issuecomment-468238875<br>
&lt;dael> astearns: Got feedback maybe resolution wasn't best<br>
&lt;dael> myles: Anyone on the call not at F2F and has opinions?<br>
&lt;dael> chris: I wasn't. I agree most impl do full unicode case folding and ascii only case folding will give odd results. Myles you said WK hasn't adhered<br>
&lt;dael> myles: True on iOS and OSX<br>
&lt;dael> chris: You sure it's a big performance regression? People are saying it wasn't<br>
&lt;dael> myles: Haven't measured b/c haven't impl.<br>
&lt;dael> chris: But you do have the function call<br>
&lt;dael> florian: At the F2F metioned the other function is not called anywhere in WK codebase<br>
&lt;dael> myles: I don't have anything new that I didn't bring up. We've never had behavior and never had compat problems. Doesn't seem need to slow down browser<br>
&lt;dael> AmeliaBR: Main new information is that when we resolved it was based on other impl doing quick look and saying I think we do this but in issue people who know better say no we follow the spec and they're concerned about changing<br>
&lt;dael> fantasai: WE do have concerns about [missed] as well<br>
&lt;fantasai> s/about [missed]/from i18n/<br>
&lt;dael> chris: CSS does ascii only case folding for syntax but for this the font name can be anything and is usually localized. but most scripts don't have two cases<br>
&lt;dael> florian: Mostly cyrillic and greek?<br>
&lt;dael> fantasai: No because accented latin is also outside ascii<br>
&lt;dael> florian: I think in practice concern will be less accented Latin then Cyrillic<br>
&lt;dael> florian: It's not that it will not happen, just more common to have problems in Cyrillic. Could be in all sorts of places<br>
&lt;dael> astearns: Interesting there haven't been reported compat issues given this incompat in code bases. I wonder if people modify to work around this<br>
&lt;dael> chris: Could be[missed]<br>
&lt;dael> dbaron: People use diff fonts on diff platofrms so there may be compat issue on other platforms<br>
&lt;dael> myles: Would other browsers be interested in speeding this up in their impl?<br>
&lt;dael> [silence]<br>
&lt;dael> astearns: Given issue comments I doubt this is the case.<br>
&lt;chris> s/[missed]/, it was interesting tht Android matches on a Latin-only xml font database, so people munge the fonts names to Latin there<br>
&lt;dael> fantasai: Given no one is seriously concerned about perf. We have existing impl that are aligned and it's the right thing to do from i18n view it seems we should not make this change and keep spec as is with full unicode case folding<br>
&lt;dael> florian: Agree and at same time I would also understand if Apple didn't prioritize this<br>
&lt;dbaron> I agree with fantasai as well.<br>
&lt;dael> myles: Like I said we don't have compat issues<br>
&lt;dael> fantasai: If we had interop on ascii only and there was perf concerns I would understand. But apparently neither is true so we should do the right thing<br>
&lt;dael> chris: I agree. It would be interesting to know where there are compat, like actual fonts used on the web that give different results without the unicode case folding. WE can punt that to i18n and request examples<br>
&lt;dael> myles: Not sure if it's worth it, but we can do it<br>
&lt;dael> chris: If there are no examples then it doesn't matter and impl might want to use simplier. If there are examples it might persuade you down the line where we didn't realize we were disadvantaging someone. Maybe there is a community with a problem, maybe there isn't<br>
&lt;dael> astearns: myles are you okay reverting previous and close no change?<br>
&lt;dael> myles: yes<br>
&lt;dael> astearns: obj to Void the previous resolution and close no change?<br>
&lt;dael> RESOLVED: Void the previous resolution and close no change<br>
&lt;dael> astearns: If new information comes up we can look at it<br>
</details>


-- 
GitHub Notification of comment by css-meeting-bot
Please view or discuss this issue at https://github.com/w3c/csswg-drafts/issues/3675#issuecomment-472494229 using your GitHub account

Received on Wednesday, 13 March 2019 16:16:33 UTC