Bug 201575 – Option + arrow moves caret past whitespace on iOS

• Component: HTML Editing
• Status: REOPENED

[Ryosuke Niwa]

See https://bugs.webkit.org/show_bug.cgi?id=201531.

This bug tracks the fix for the hardware keyboard.

<rdar://problem/51404664>

[Ryosuke Niwa]

Created attachment 378278 [details]
Fixes the bug

[Ryosuke Niwa]

Created attachment 378279 [details]
Fixed non-iOS builds

[Ryosuke Niwa]

Created attachment 378282 [details]
GTK build fix attempt

[Ryosuke Niwa]

Created attachment 378283 [details]
Another GTK build fix attempt

[Ryosuke Niwa]

Created attachment 378284 [details]
Yet another GTK build fix attempt

[Ryosuke Niwa]

Created attachment 378399 [details]
Fixed build more

[Wenson Hsieh]

Comment on attachment 378399 [details]
Fixed build more

View in context: https://bugs.webkit.org/attachment.cgi?id=378399&action=review

> Source/WebCore/ChangeLog:16
> +        are involed from author scripts.

Nit - involed => involved

> Source/WebCore/editing/VisibleUnits.cpp:772
> +    return findNextWordFromIndex(text, offset, NextWordDirection::Backward, iosMode);

Nit - maybe name the variable nextWordMode or whitespaceMode instead of iosMode? (Or perhaps more verbose — nextWordModeInIOS).

> Source/WebCore/editing/VisibleUnits.h:45
> +WEBCORE_EXPORT VisiblePosition nextWordPosition(const VisiblePosition &, NextWordModeInIOS = NextWordModeInIOS::LegacyStopBeforeWord);

Nit - might consider removing the extra space before the ampersand in `const VisiblePosition &`, while we’re touching this line.

> Source/WebCore/platform/text/TextBoundaries.h:51
> +    enum class NextWordDirection { Forward, Backward };

I usually use `enum class NextWordDirection : bool { ~ }` for these, but it doesn’t really matter since NextWordDirection isn’t stored anywhere.

> Source/WebCore/platform/text/TextBoundaries.h:52
> +    enum class NextWordModeInIOS { LegacyStopBeforeWord, StopAfterWord };

(Ditto with the bool enum class)

[Ryosuke Niwa]

Thanks for the review.

(In reply to Wenson Hsieh from comment #7)
> Comment on attachment 378399 [details]
> Fixed build more
> 
> View in context:
> https://bugs.webkit.org/attachment.cgi?id=378399&action=review
> 
> > Source/WebCore/ChangeLog:16
> > +        are involed from author scripts.
> 
> Nit - involed => involved

Fixed.

> > Source/WebCore/editing/VisibleUnits.cpp:772
> > +    return findNextWordFromIndex(text, offset, NextWordDirection::Backward, iosMode);
> 
> Nit - maybe name the variable nextWordMode or whitespaceMode instead of
> iosMode? (Or perhaps more verbose — nextWordModeInIOS).

Will use nextWordModeInIOS. whitespaceMode isn't right because we consider delimiters such as "," as well.

> > Source/WebCore/editing/VisibleUnits.h:45
> > +WEBCORE_EXPORT VisiblePosition nextWordPosition(const VisiblePosition &, NextWordModeInIOS = NextWordModeInIOS::LegacyStopBeforeWord);
> 
> Nit - might consider removing the extra space before the ampersand in `const
> VisiblePosition &`, while we’re touching this line.

Sure, done.

> > Source/WebCore/platform/text/TextBoundaries.h:51
> > +    enum class NextWordDirection { Forward, Backward };
> 
> I usually use `enum class NextWordDirection : bool { ~ }` for these, but it
> doesn’t really matter since NextWordDirection isn’t stored anywhere.

Thanks for catching that. I always forget to specify the enum class' width. I've been coding too long in old C/C++ :(

> > Source/WebCore/platform/text/TextBoundaries.h:52
> > +    enum class NextWordModeInIOS { LegacyStopBeforeWord, StopAfterWord };
> 
> (Ditto with the bool enum class)

Fixed this one too.

[Ryosuke Niwa]

Created attachment 378444 [details]
Patch for landing

[Ryosuke Niwa]

Comment on attachment 378444 [details]
Patch for landing

Wait for EWS again.

[Ryosuke Niwa]

Finally!

[Ryosuke Niwa]

Comment on attachment 378444 [details]
Patch for landing

Ugh... wrong + :(

[WebKit Commit Bot]

Comment on attachment 378444 [details]
Patch for landing

Clearing flags on attachment: 378444

Committed r249709: <https://trac.webkit.org/changeset/249709>

[WebKit Commit Bot]

All reviewed patches have been landed.  Closing bug.

[Russell Epstein]

Reverted r249709 for reason:

Layout test added in this patch has been consistently failing since it landed.

Committed r249838: <https://trac.webkit.org/changeset/249838>