415166 - Left table column is wider than it used to be

Status: REOPENED

(Core :: Layout: Tables, defect)

Description

[Martijn Wargers (dead)]

Attachment: testcase

This regressed when the reflow branch landed.

The left table column is now wider than it used to be.
Older builds (like branch), seem to do more like what IE7 is doing also.

Comment 1

[Daniel Holbert [:dholbert]]

This seems like a bug in table-width-balancing.

As I resize the window to be wider, Trunk gives extra width to the first column.

But I don't know why we'd do that, when we've already satisfied it's preferred width and haven't satisfied the second column's preferred width.  It seems like the correct behavior would be to leave first column at its preferred width, until the second column also gets its preferred width.  (This seems to be what FF2 does)

Comment 2

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

We've already satisfied the preferred width of both columns.

Does IE change if you switch the order of the rows?

Comment 3

[Daniel Holbert [:dholbert]]

(In reply to comment #2)
> We've already satisfied the preferred width of both columns.

Ah, ok -- I was confused because the JS tweak in the testcase changes the col-widths slightly, at least on my machine.  (probably due to the text insertion)

Without the JS tweak, the columns hit 129 px & 855 px (their pref widths) simultaneously.

Comment 4

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

Does the JS change that?  If so, then there probably is a bug.

Comment 5

[Daniel Holbert [:dholbert]]

Attachment: testcase 2a (quirks mode)

You can ignore comment #3 -- I think I was confused at the time because the cellspacing/cellpadding was messing with the widths a bit, or something.  I can't get the same behavior I was talking about then.

In any case, I think this new testcase demonstrates what's going on here.

In Firefox 2, the first column stays skinny, and all excess width goes to the second column.

In Firefox 3 (and current mozilla-central), excess width is distributed roughly equally between the columns.

In IE8, the first row ordering behaves like FF2 and the second behaves like FF3.

Comment 6

[Daniel Holbert [:dholbert]]

Attachment: testcase 2b (standards mode)

Comment 7

[Daniel Holbert [:dholbert]]

When I force standards mode (as in testcase 2b), IE8 & FF2 switch to behave like FF3, in both row-orderings.

AFAICT, the only regression here is that we've removed one quirks-mode behavior -- a behavior which IE itself only exhibits in one of the two possible row orderings.

So, I think this should probably be resolved as WONTFIX or INVALID.

Comment 8

[Daniel Holbert [:dholbert]]

Attachment: original testcase, with rows reversed

Just to give further support to my previous comment -- here's the original testcase with the rows reversed.

IE8 matches FF3 on this testcase (thereby matching FF3's original-testcase-behavior).  That is to say, the first column grows with the width of the window.

(Compare that to how IE handled the original testcase -- the first column's width was fixed there, with all excess width getting allocated to the second column.)

Comment 9

[Guilherme Lima]

Is this and https://bugzilla.mozilla.org/show_bug.cgi?id=477004 the same bug?
IE10 and Chrome 30 display the same thing on all 4 testcases.
Firefox only displays the same thing as them on testcase 2b (standards mode)

Comment 10

[Justin [:JW_SoftvisionQA]]

(In reply to Guilherme Lima from comment #9)
> IE10 and Chrome 30 display the same thing on all 4 testcases.
> Firefox only displays the same thing as them on testcase 2b (standards mode)

This is still the same result on this issue.