RESOLVED FIXED 198763
[Text autosizing] [iPadOS] Revise our heuristics to determine idempotent text autosizing candidates
https://bugs.webkit.org/show_bug.cgi?id=198763
Summary [Text autosizing] [iPadOS] Revise our heuristics to determine idempotent text...
Wenson Hsieh
Reported 2019-06-11 12:54:33 PDT Comment hidden (obsolete)
Attachments
WIP patch (11.74 KB, patch)
2019-06-12 08:56 PDT, Wenson Hsieh
no flags
Patch (17.32 KB, patch)
2019-06-24 15:34 PDT, Wenson Hsieh
no flags
Less code, more words (19.41 KB, patch)
2019-06-24 17:21 PDT, Wenson Hsieh
simon.fraser: review+
Patch for landing (23.16 KB, patch)
2019-06-24 20:24 PDT, Wenson Hsieh
no flags
Wenson Hsieh
Comment 1 2019-06-12 08:56:16 PDT Comment hidden (obsolete)
Wenson Hsieh
Comment 2 2019-06-24 14:22:16 PDT
Wenson Hsieh
Comment 3 2019-06-24 15:34:12 PDT Comment hidden (obsolete)
Wenson Hsieh
Comment 4 2019-06-24 16:05:47 PDT
Comment on attachment 372807 [details] Patch I’m going to see if I can make this a bit less crazy.
Simon Fraser (smfr)
Comment 5 2019-06-24 16:18:39 PDT
Comment on attachment 372807 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=372807&action=review > Source/WebCore/rendering/style/RenderStyle.cpp:495 > +bool RenderStyle::isIdempotentTextAutosizingCandidate() const You def. need some words about how this code came into being, and ideally you'd commit the tools used to generate this code, and the training dataset.
Wenson Hsieh
Comment 6 2019-06-24 16:32:05 PDT
(In reply to Simon Fraser (smfr) from comment #5) > Comment on attachment 372807 [details] > Patch > > View in context: > https://bugs.webkit.org/attachment.cgi?id=372807&action=review > > > Source/WebCore/rendering/style/RenderStyle.cpp:495 > > +bool RenderStyle::isIdempotentTextAutosizingCandidate() const > > You def. need some words about how this code came into being, Yes, will do! I put some words in the overall WebCore ChangeLog entry, but this definitely warrants more detail. > and ideally > you'd commit the tools used to generate this code, and the training dataset. Interesting...into open source, that is? I think we’ll need to check if that’s kosher :/
Simon Fraser (smfr)
Comment 7 2019-06-24 16:43:06 PDT
Maybe not, but to be maintainable, we have to be able to re-run your tools at any time.
Wenson Hsieh
Comment 8 2019-06-24 16:46:47 PDT
(In reply to Simon Fraser (smfr) from comment #7) > Maybe not, but to be maintainable, we have to be able to re-run your tools > at any time. Indeed. They are already accessible to all of us (WebKit/Safari at Apple), but I’ll work towards getting the tools into the main Safari internal repository.
Wenson Hsieh
Comment 9 2019-06-24 17:21:39 PDT
Created attachment 372811 [details] Less code, more words
Simon Fraser (smfr)
Comment 10 2019-06-24 17:33:01 PDT
Comment on attachment 372811 [details] Less code, more words View in context: https://bugs.webkit.org/attachment.cgi?id=372811&action=review > Source/WebCore/ChangeLog:17 > + Test: fast/text-autosizing/ios/idempotentmode/idempotent-autosizing-skip.html Be careful with "skip" in a test name (I think we have a convention that filenames with "skipped" in them are skipped). I would call this something else. > Source/WebCore/rendering/style/RenderStyle.cpp:497 > + auto fields = OptionSet<AutosizeStatus::Fields>::fromRaw(m_inheritedFlags.autosizeStatus); Need a comment at the top of this function referencing the radar that can point someone to the tools used to generate this code.
Wenson Hsieh
Comment 11 2019-06-24 20:24:18 PDT
Created attachment 372817 [details] Patch for landing
WebKit Commit Bot
Comment 12 2019-06-24 21:06:58 PDT
Comment on attachment 372817 [details] Patch for landing Clearing flags on attachment: 372817 Committed r246781: <https://trac.webkit.org/changeset/246781>
Note You need to log in before you can comment on or make changes to this bug.