Fix table sizing when 'max-width' is used

Fix table sizing when 'max-width' is used

r143534 make <table> abide by 'max-width' all the time which is wrong.
Per the CSS specification, a table should be wide enough to fit its
content, regardless of 'max-width'.

r140479 fixed part of the regression from that change but made the
same fatal mistake by constraining min-content to fit 'max-width'.

The fix is to avoid constraining min-content and ensure that the table
logical width is at least its min-content size.

BUG=173966
TEST=fast/table/html-table-width-max-width-constrained.html

Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=149011

[esprehn, 2013-04-19 08:07:25]

All the try jobs are red, were those unrelated? Also does Firefox pass that
test?

https://codereview.chromium.org/14141004/diff/1/LayoutTests/fast/table/css-ta...
File LayoutTests/fast/table/css-table-max-width-expected.txt (right):

https://codereview.chromium.org/14141004/diff/1/LayoutTests/fast/table/css-ta...
LayoutTests/fast/table/css-table-max-width-expected.txt:13: PASS
maxWidthZeroAutoLayout.getBoundingClientRect().width is 113
Funny this output was wrong.

https://codereview.chromium.org/14141004/diff/1/Source/core/rendering/RenderT...
File Source/core/rendering/RenderTable.cpp (right):

https://codereview.chromium.org/14141004/diff/1/Source/core/rendering/RenderT...
Source/core/rendering/RenderTable.cpp:314: ASSERT(logicalWidth() >=
minPreferredLogicalWidth());
Yay!

[Julien - ping for review, 2013-04-19 21:29:12]

> All the try jobs are red, were those unrelated?

Mostly. I see the existing test failing on one platform and I think it's because
we need to use Ahem to ensure consistent sizing across platforms.

> Also does Firefox pass that test?

Yes, both new tests are consistent with Firefox (we don't totally match their
sizing though but our behaviors are matching).

https://codereview.chromium.org/14141004/diff/1/LayoutTests/fast/table/css-ta...
File LayoutTests/fast/table/css-table-max-width-expected.txt (right):

https://codereview.chromium.org/14141004/diff/1/LayoutTests/fast/table/css-ta...
LayoutTests/fast/table/css-table-max-width-expected.txt:13: PASS
maxWidthZeroAutoLayout.getBoundingClientRect().width is 113
On 2013/04/19 08:07:25, esprehn wrote:
> Funny this output was wrong.

Yeah, I didn't double check the test on other browsers and this was actually a
difference with Firefox :-/

[ojan, 2013-04-19 22:44:01]

lgtm

I think this is fine for now. Long-term I've been hoping that we add the
following to html.css:
table {
    min-width: min-content;
}

Then, if pages want a different min-width value they can set it. I'm not sure if
there's already too much content that depends on having both a min-width set and
expecting the min-width to be min-content anyways.

[ojan, 2013-04-19 22:44:30]

The thing I would like about that is that it'd be one few way in which tables
would be a special snowflake.

[commit-bot: I haz the power, 2013-04-24 16:03:41]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/jchaffraix@chromium.org/14141004/11001

[commit-bot: I haz the power, 2013-04-24 16:04:04]

Change committed as 149011