Bug 79853 – [EFL][DRT] WebKit/DRT should not paint webview for the pixel tests.

• Status: RESOLVED
• Resolution: WONTFIX
• Component: WebKit EFL

[Ryuan Choi]

For now, some results of pixel test are wrong.
(For example, LayoutTests/fast/events/autoscroll.html)

In order to fix this, I suggest capturing Ecore_Evas instead of painting webview.

[Ryuan Choi]

Created attachment 129368 [details]
Patch

[Ryuan Choi]

Comment on attachment 129368 [details]
Patch

clear flags until I check how to rebase test cases.

[Ryuan Choi]

Comment on attachment 129368 [details]
Patch

I checked png results which is newly failed.

Some test cases are because of scrollbar.
Although mockScrollbar is enabled, WebKit/Efl create native scrollbar.
I will create new bug to fix them.

The other test case is fast/forms/textarea-placeholder-set-attribute.html
and it can be reproduce on EWeblauncher. (not always. 50% reproduced)

So, I believe that patch is valid for bug.

Could you review this?

[Ryuan Choi]

Comment on attachment 129368 [details]
Patch

I realize that DRT/Efl(tiled) can not render scrolled contents with this patch correctly.
(ex - compositing/geometry/fixed-position.html)

So I cleared flags until I found proper way.

BTW

[Ryuan Choi]

(In reply to comment #4)
> (From update of attachment 129368 [details])
> I realize that DRT/Efl(tiled) can not render scrolled contents with this patch correctly.
> (ex - compositing/geometry/fixed-position.html)

I created Bug 84933 for compositing/geometry/fixed-position.html

[Raphael Kubo da Costa (:rakuco)]

Comment on attachment 129368 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=129368&action=review

Looks good API-wise, just make sure it doesn't cause test regressions.

> Tools/DumpRenderTree/efl/PixelDumpSupportEfl.cpp:53
> +    RefPtr<cairo_surface_t> viewSurface = adoptRef(cairo_image_surface_create_for_data((unsigned char*)pixels, CAIRO_FORMAT_ARGB32, width, height, width * 4));

Please use a C++-style cast here.

[Ryuan Choi]

Created attachment 139416 [details]
Patch

[Ryuan Choi]

(In reply to comment #6)
> (From update of attachment 129368 [details])
> View in context: https://bugs.webkit.org/attachment.cgi?id=129368&action=review
> 
> Looks good API-wise, just make sure it doesn't cause test regressions.
> 

Thanks.

There are no regression in my local.
(compositing/geometry/fixed-position.html is in Skipped now.)

> > Tools/DumpRenderTree/efl/PixelDumpSupportEfl.cpp:53
> > +    RefPtr<cairo_surface_t> viewSurface = adoptRef(cairo_image_surface_create_for_data((unsigned char*)pixels, CAIRO_FORMAT_ARGB32, width, height, width * 4));
> 
> Please use a C++-style cast here.

Done.

[Chang Shu]

Comment on attachment 139416 [details]
Patch

The comments in 81315 mentioned this bug. Does one depend on the other?

[Raphael Kubo da Costa (:rakuco)]

(In reply to comment #9)
> (From update of attachment 139416 [details])
> The comments in 81315 mentioned this bug. Does one depend on the other?

I believe so.

[Eric Seidel (no email)]

Comment on attachment 139416 [details]
Patch

I'm happy to rubber-stamp this, but I"m not very useful for a real review.

[WebKit Review Bot]

Comment on attachment 139416 [details]
Patch

Clearing flags on attachment: 139416

Committed r116461: <http://trac.webkit.org/changeset/116461>

[WebKit Review Bot]

All reviewed patches have been landed.  Closing bug.

[Raphael Kubo da Costa (:rakuco)]

Reopening to get some feedback. This commit appears to have broken the painting of scrollbars with both backing stores when running run-webkit-tests with --pixel-tests. Try running RWT on fast/box-sizing, for example.

[Ryuan Choi]

(In reply to comment #14)
> Reopening to get some feedback. This commit appears to have broken the painting of scrollbars with both backing stores when running run-webkit-tests with --pixel-tests. Try running RWT on fast/box-sizing, for example.

I checked and reproduces it.

Now, I realized that backing stores can not render mock scrollbars because they call Frame::paintContents instead of ScrollView::paint.

Sorry, I will find a way to fix it but it will take more time.

[Raphael Kubo da Costa (:rakuco)]

I will revert this for now then, we can't stay with the buggy code.

[Raphael Kubo da Costa (:rakuco)]

Reverted in <http://trac.webkit.org/changeset/117879>.

[Ryuan Choi]

ewebkit1 was dropped.