[webkit-dev] Simplifying syntax in test_expectations.txt (bug 86691)

Ryosuke Niwa rniwa at webkit.org
Thu May 17 12:19:35 PDT 2012

On Thu, May 17, 2012 at 11:54 AM, Maciej Stachowiak <mjs at apple.com> wrote:
> On May 17, 2012, at 11:28 AM, Ryosuke Niwa <rniwa at webkit.org> wrote:
> On Thu, May 17, 2012 at 11:08 AM, Peter Kasting <pkasting at google.com>wrote:
>> On Thu, May 17, 2012 at 9:26 AM, Dimitri Glazkov <dglazkov at chromium.org>wrote:
>>> I actually quite like
>>> the clear delineation between test modifiers and test expectations.
>> Me too.  Please please please leave them on opposite sides.  All these
>> proposals that try to put them both before the test name are much harder
>> for me to understand.  Even dropping the all-caps format for keywords is a
>> loss for me as this differentiation makes it easier for me to scan the line
>> and distinguish the three important pieces of a line: what, where, and
>> how.
>> I find most of the changes that have been proposed so far to be a
>> significant decrease in clarity, except for the idea to change BUGXX###
>> into links (for all bug trackers, not just non-WebKit ones), which I think
>> would be fine; and the idea of moving SLOW, SKIP, and WONTFIX from the left
>> side to the right, as these seem more like "expectations about how the test
>> will run" than "platforms on which the test fails".
> Interesting. I find the new format (I've replaced [] by () since that
> worked better for me) much easier to read than the old format because now
> all meta data is in one place. For me, I'm most annoyed by the unreasonably
> long bug URLs in the new format. Maybe putting the URL isn't such a great
> idea after all :(
> Perhaps aligning the fields column after the bug URL would improve
> readability (though it makes things a little harder to edit):
> crbug.com/24182    Slow (Mac Debug)
> fast/css/large-list-of-rules-crash.html
> crbug.com/24182    Slow (Linux Mac Debug)
> fast/dom/Window/window-postmessage-clone-really-deep-array.html
> crbug.com/24182    Slow (Mac Debug)
> fast/forms/select-set-length-with-mutation-remove.html
> crbug.com/24182    Slow (All)             fast/js/regexp-overflow.html
> crbug.com/24182    Slow (Debug)
> fast/js/toString-and-valueOf-override.html
> crbug.com/24182    Slow (Debug)           html5lib/webkit-resumer.html
> crbug.com/24182    Slow (Win Release)     http/tests/loading/
> onload-vs-immediate-refresh.pl
> crbug.com/79910    Image (SnowLeopard)
>  fast/text/international/bidi-linebreak-001.html
> crbug.com/79910    Image (SnowLeopard)
>  fast/text/international/bidi-linebreak-002.html
> crbug.com/79910    Image (SnowLeopard)
>  fast/text/international/bidi-linebreak-003.html
> webkit.org/b/58924 Pass Timeout (Mac)
> plugins/mouse-click-iframe-to-plugin.html
> webkit.org/b/60099 Pass Crash (Debug)
> fast/dom/Attr/access-after-element-destruction.html
> webkit.org/b/60097 Pass Text (Debug)
>  fast/dom/HTMLLinkElement/link-and-subresource-test.html
We certainly could. We treat \s+ as \s anyway so we have a lot of freedom
here. However, aligning things using whitespace in test_expectations.txt is
just as problematic in C++ code because the modifiers and platforms can
e.g. Pass Image+Text Image Crash Timeout (SnowLeopard Leopard Linux Win7

> I would also suggest omitting (All), particularly if it turns out to be
> the common case.

Yes, a lot of entries tend to be (All) so we should make it implicit
when omitted or not have "All" at all (I prefer not having it at all).

- Ryosuke
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.webkit.org/pipermail/webkit-dev/attachments/20120517/ed904009/attachment.html>

More information about the webkit-dev mailing list