[webkit-reviews] review requested: [Bug 21407] Add a number of new layout tests, all passing on ToT : [Attachment 24373] Improved tests + new results

[bugzilla-daemon at webkit.org]

Pam Greene <pam at chromium.org> has asked  for review:
Bug 21407: Add a number of new layout tests, all passing on ToT
https://bugs.webkit.org/show_bug.cgi?id=21407

Attachment 24373: Improved tests + new results
https://bugs.webkit.org/attachment.cgi?id=24373&action=edit

------- Additional Comments from Pam Greene <pam at chromium.org>
Some of the issues raised for improving these tests are intractable at this
point, but I did what I could.

All: wrote more informative descriptions about what's being tested and what to
expect when looking at the results.

fast/css/border-height.html: this was a painting issue related to rounding in
the graphics engine, so unfortunately only the pixel test is descriptive.

fast/dom/Window/orphaned-frame-access.html: improved the output

fast/forms/image-border.html: I can't tell for sure, but I believe this was
also a painting issue.

fast/forms/input-field-text-truncated.html: unfortunately at this late stage I
don't know whether it was a layout bug or a painting problem, so I don't know
whether the render tree alone would catch the error.  Likewise, although I
agree that using "em" is odd here, looking back at the bug report it was a very
fiddly edge case that caused the (rounding error) bug to show up, so I'm
reluctant to change anything about the font sizes for fear of making the test
entirely useless.

fast/forms/input-type-text-min-width.html: the result here will be affected by
whatever decision is made in bug 15312, but following the philosophy in bug
21419, it's worth adding the test now to catch unintentional changes anyway.

fast/forms/textarea-width.html: ditto

fast/parser/bad-xml-entity.xml: renamed from .html, and removed the encoding
specification. The XML parser, unlike the HTML parser, produces a parsing error
and would never see a dumpAsText() directive (I tried), so it's stuck as a
render-tree test now.

http/tests/misc/crash-multiple-family-fontface.html: moved from
http/tests/loading/

-- Tests removed from the patch:

fast/dom/DOMImplementation/singleton-modifications.html: I don't fully
understand what this is testing, and I'm not yet convinced it checks what it
intends to.  I've asked the original author for clarification and will post it
in a separate bug later.

fast/text/international/ideo_punc.html: this is a dupe of
fast/text/line-breaks-after-ideographic-comma-or-full-stop.html.