[webkit-dev] Making PopupMenuClient testable

[Johnny Ding]

Fady, please see my comments inline, Thank.

在 2011年10月10日 下午10:57，Fady Samuel <fsamuel at chromium.org>写道：

> +darin
>
> Johnny, I'm not sure what you mean. By the render dump, do you mean the
> pixel dump? Popups don't show up in pixel dumps. I'm not sure what
> information you can get from that.
>
> What I mean is regular render tree dump, not pixel dump, not text dump.
Please refer here<http://trac.webkit.org/browser/trunk/LayoutTests/platform/chromium-win/fast/forms/menulist-clip-expected.txt>for
test expectation and see in
TestShell::dump<http://trac.webkit.org/browser/trunk/Tools/DumpRenderTree/chromium/TestShell.cpp#L474>in
DRT chromium port



> The bug that I'm trying to address, is, among other things, the bounding
> box rect of the PopupMenuClient that is passed into the platform layer to
> show the popup does not take transforms into account.
>
After landing https://bugs.webkit.org/show_bug.cgi?id=66062, will
the PopupMenuClient take transforms into account?

 My solution in 69631 is to expose the bounding box rect through the
> PopupMenuClient interface and make it testable through window.internals.
> I've also included a layout test to show how one would test this. Right now
> the test fails, but once 66062 is applied the test will pass.
>
> Is this a reasonable approach?
>
> Thanks,
>
> Fady
>
>
> On Mon, Oct 10, 2011 at 10:41 AM, Johnny Ding <jnd at google.com> wrote:
>
>>
>>
>> 在 2011年10月7日 下午11:56，Fady Samuel <fsamuel at chromium.org>写道：
>>
>> Hi all,
>>>
>>> I've just uploaded a patch https://bugs.webkit.org/show_bug.cgi?id=69631 that's
>>> intended to make PopupMenuClient testable.
>>>
>>> On certain Chromium platforms, the position of the <select> element's
>>> popup is incorrect if the page's scale factor is not 1.0.
>>>
>>> The fix to the problem is simple, see
>>> https://bugs.webkit.org/show_bug.cgi?id=66062 , but since popups are
>>> external to WebKit on most platforms, this is difficult to test.
>>>
>>> The patch above in bug 69631 attempts to addresses this, albeit in a less
>>> than ideal way. I'm not sure what's the best way to test this, but this is
>>> my stab at it.
>>>
>>  Even if  the popup is external, the render dump (not text dump) still
>> contains layout information. You can write a layout test with page's scale
>> factor not 1.0, then generate the expectation results among WebKit ports.
>>  Do you think Is that enough for your test?
>>
>>>
>>> I'd really appreciate any comments or suggestions you can provide on my
>>> patch for bug 69631.
>>>
>>> Thanks,
>>>
>>> Fady
>>>
>>> _______________________________________________
>>> webkit-dev mailing list
>>> webkit-dev at lists.webkit.org
>>> http://lists.webkit.org/mailman/listinfo.cgi/webkit-dev
>>>
>>>
>>
>>
>> --
>> Best regards,
>> Johnny(Jianning) Ding | Software Engineer | jnd at google.com |
>>  (86)189-1186-0082
>>
>>
>


-- 
Best regards,
Johnny(Jianning) Ding | Software Engineer | jnd at google.com |
 (86)189-1186-0082
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.webkit.org/pipermail/webkit-dev/attachments/20111011/504e51d3/attachment.html>