Bug 186472 – Link drag image is inconsistently unreadable in dark mode

• Status: RESOLVED
• Resolution: FIXED
• Component: New Bugs

[Tim Horton]

Link drag image is inconsistently unreadable in dark mode

[Tim Horton]

Created attachment 342382 [details]
Patch

[EWS Watchlist]

Comment on attachment 342382 [details]
Patch

Attachment 342382 [details] did not pass win-ews (win):
Output: http://webkit-queues.webkit.org/results/8110746

New failing tests:
http/tests/security/contentSecurityPolicy/userAgentShadowDOM/allow-audio.html

[EWS Watchlist]

Created attachment 342383 [details]
Archive of layout-test-results from ews201 for win-future

The attached test failures were seen while running run-webkit-tests on the win-ews.
Bot: ews201  Port: win-future  Platform: CYGWIN_NT-6.1-2.9.0-0.318-5-3-x86_64-64bit

[Megan Gardner]

Comment on attachment 342382 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=342382&action=review

I’m nervous about just removing the check in default appearance. Can we write a test for this?

> Source/WebKit/UIProcess/Cocoa/WebViewImpl.mm:5019
>      NSAppearanceName appearance = [[m_view effectiveAppearance] bestMatchFromAppearancesWithNames:@[ NSAppearanceNameAqua, NSAppearanceNameDarkAqua ]];

Are we sure that we can do this without leaking dark mode to parts where we are sure we *don’t* want it to be used?

> Source/WebKitLegacy/mac/WebView/WebView.mm:5281
>      NSAppearanceName appearance = [[self effectiveAppearance] bestMatchFromAppearancesWithNames:@[ NSAppearanceNameAqua, NSAppearanceNameDarkAqua ]];

Ditto

[Timothy Hatcher]

Comment on attachment 342382 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=342382&action=review

> Source/WebCore/platform/mac/DragImageMac.mm:306
> +    LocalDefaultSystemAppearance localAppearance(true, page ? page->defaultAppearance() : true);

We need a better way to allow getting the true effective appearance for places like this that need it. This does not feel right as-is, considering the changes to defaultAppearance().

>> Source/WebKit/UIProcess/Cocoa/WebViewImpl.mm:5019
>>      NSAppearanceName appearance = [[m_view effectiveAppearance] bestMatchFromAppearancesWithNames:@[ NSAppearanceNameAqua, NSAppearanceNameDarkAqua ]];
> 
> Are we sure that we can do this without leaking dark mode to parts where we are sure we *don’t* want it to be used?

Megan is right, we cannot do this here. It will cause Safari to use dark mode colors and form controls for pages.

[Timothy Hatcher]

Comment on attachment 342382 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=342382&action=review

>> Source/WebCore/platform/mac/DragImageMac.mm:306
>> +    LocalDefaultSystemAppearance localAppearance(true, page ? page->defaultAppearance() : true);
> 
> We need a better way to allow getting the true effective appearance for places like this that need it. This does not feel right as-is, considering the changes to defaultAppearance().

I think this is fine now. Sorry for the confusion. I think I want to make a patch that flips defaultAppearance to be prefersDarkInterface, so it is easier to read everywhere.

>>> Source/WebKit/UIProcess/Cocoa/WebViewImpl.mm:5019
>>>      NSAppearanceName appearance = [[m_view effectiveAppearance] bestMatchFromAppearancesWithNames:@[ NSAppearanceNameAqua, NSAppearanceNameDarkAqua ]];
>> 
>> Are we sure that we can do this without leaking dark mode to parts where we are sure we *don’t* want it to be used?
> 
> Megan is right, we cannot do this here. It will cause Safari to use dark mode colors and form controls for pages.

Actually, I think this is fine now! I forgot LocalDefaultSystemAppearance now checks for both useSystemAppearance and useDefaultAppearance. In the Safari case, useSystemAppearance will still be false.

[Tim Horton]

(In reply to Timothy Hatcher from comment #6)
> Comment on attachment 342382 [details]
> Patch
> 
> View in context:
> https://bugs.webkit.org/attachment.cgi?id=342382&action=review
> 
> >> Source/WebCore/platform/mac/DragImageMac.mm:306
> >> +    LocalDefaultSystemAppearance localAppearance(true, page ? page->defaultAppearance() : true);
> > 
> > We need a better way to allow getting the true effective appearance for places like this that need it. This does not feel right as-is, considering the changes to defaultAppearance().
> 
> I think this is fine now. Sorry for the confusion. I think I want to make a
> patch that flips defaultAppearance to be prefersDarkInterface, so it is
> easier to read everywhere.

Yes please!

> >>> Source/WebKit/UIProcess/Cocoa/WebViewImpl.mm:5019
> >>>      NSAppearanceName appearance = [[m_view effectiveAppearance] bestMatchFromAppearancesWithNames:@[ NSAppearanceNameAqua, NSAppearanceNameDarkAqua ]];
> >> 
> >> Are we sure that we can do this without leaking dark mode to parts where we are sure we *don’t* want it to be used?
> > 
> > Megan is right, we cannot do this here. It will cause Safari to use dark mode colors and form controls for pages.
> 
> Actually, I think this is fine now! I forgot LocalDefaultSystemAppearance
> now checks for both useSystemAppearance and useDefaultAppearance. In the
> Safari case, useSystemAppearance will still be false.

Yeah, that’s what I was banking on. OK!

[WebKit Commit Bot]

Comment on attachment 342382 [details]
Patch

Clearing flags on attachment: 342382

Committed r232731: <https://trac.webkit.org/changeset/232731>

[WebKit Commit Bot]

All reviewed patches have been landed.  Closing bug.

[Radar WebKit Bug Importer]

<rdar://problem/41019148>