333352 - remove support for proportional widths on <col>

Status: RESOLVED FIXED

(Core :: Layout: Tables, defect)

Description

[David Baron :dbaron: (⌚️UTC-5, no longer working on Mozilla)]

We support proportional widths (e.g., "0*", "1*", "2*", "*") on <col> elements in HTML tables.  I haven't been able to find any other browser that supports them, and the spec that defines them, http://www.w3.org/TR/html4/struct/tables.html#h-11.2.4.4 , defines percentage widths incorrectly (from reality) and defines proportional widths a lot like the way percentage widths really work.  (That said, there are probably some differences in implementation.)

I don't see any reason to continue supporting this feature.  Does anyone else?

I propose to remove the content-side of the support for it in this bug and remove the layout-side on the reflow branch.

Comment 1

[David Baron :dbaron: (⌚️UTC-5, no longer working on Mozilla)]

Attachment: testcase probing for support for proportional widths on <col>

Comment 2

[Bernd]

Thats from the time when supporting HTML 4.0 was a primary goal. It hurts to see my time going to the trash bin.

It seems that opera, safari and konq do allready have issues with the width at all. (http://www.robinlionheart.com/stds/html4/results see the col-width row)

http://www.re.be/css2xslfo/1_1/release_notes is just happy to have added pcw although I am not sure that they rely on * notation.

And while we kill underutilized features rowspan="0" and colspan="0" come somehow very quick into mind.

Comment 3

[David Baron :dbaron: (⌚️UTC-5, no longer working on Mozilla)]

(In reply to comment #2)
> And while we kill underutilized features rowspan="0" and colspan="0" come
> somehow very quick into mind.

Those actually seem clearly defined and useful, though.

Comment 4

[David Baron :dbaron: (⌚️UTC-5, no longer working on Mozilla)]

Attachment: patch

I'll post cleanup of nsAttrValue::ParseSpecialIntValue as a followup patch.

Comment 5

[Boris Zbarsky [:bzbarsky]]

Comment on attachment 217802 [details] [diff] [review]
patch

Looks reasonable to me; perhaps add a comment pointing to this bug as the reason we don't implement the "HTML spec" here?

Comment 6

[Bernd]

Comment on attachment 217802 [details] [diff] [review]
patch

what boris said and I assume that we have enough time before the branch lands to get feedback.

>I'll post cleanup of nsAttrValue::ParseSpecialIntValue as a followup patch.

Isn't this code also used by the html-frames?

Comment 7

[David Baron :dbaron: (⌚️UTC-5, no longer working on Mozilla)]

Requesting blocking1.9a1 since this is a feature I plan to remove with the reflow branch, and it's very easy to get feedback on that removal in the earlier alpha, and I'd rather have the feedback, if any, sooner rather than later.

Comment 8

[Chris Knoll]

(In reply to comment #0)
> We support proportional widths (e.g., "0*", "1*", "2*", "*") on <col> elements
> in HTML tables.  I haven't been able to find any other browser that supports
> them, and the spec that defines them,
> http://www.w3.org/TR/html4/struct/tables.html#h-11.2.4.4 , defines percentage
> widths incorrectly (from reality) and defines proportional widths a lot like
> the way percentage widths really work.  (That said, there are probably some
> differences in implementation.)

I disagree with this assessment.  The spec is very clear about what you do:
Fixed and percentage widths are allocated.  The remaining * groups are calculated by adding up the number of *s you have and divide the remaining horizontal space.  Then each * portion is given to each column.

So if you have 600px horizontal space and columns 30px, 10%, 20%, 1*,2*,3*,4*, then your column widths would be:
30px, 60px, 120px, 39px, 78px, 117px, 156px

Comment 9

[David Baron :dbaron: (⌚️UTC-5, no longer working on Mozilla)]

You don't seem interested in the bit about "a lot like the way percentage widths really work" (which coincidentally a lot of the Web depends on).

Comment 10

[David Baron :dbaron: (⌚️UTC-5, no longer working on Mozilla)]

Fix checked in to trunk.

Comment 11

[David Baron :dbaron: (⌚️UTC-5, no longer working on Mozilla)]

(In reply to comment #6)
> Isn't this code also used by the html-frames?

Nope, nsHTMLFrameSetElement has more-specific parsing code.

(In reply to comment #4)
> I'll post cleanup of nsAttrValue::ParseSpecialIntValue as a followup patch.

I filed bug 341683 to do this.

Comment 12

[Chris Knoll]

(In reply to comment #9)
> You don't seem interested in the bit about "a lot like the way percentage
> widths really work" (which coincidentally a lot of the Web depends on).
> 

I thought that was what 'quirks' mode was for, and I also thought there was interest in having browser behavior follow the spec.  Otherwise, why not go back to the old way of aligning images to the bottom of the box instead of baseline?  Changing that behavior broke a lot of web pages too.

Comment 13

[David Baron :dbaron: (⌚️UTC-5, no longer working on Mozilla)]

Attachment: remove proportional units from style system

The previous patch removed the only user of proportional units in the style system; this removes the units.

Comment 14

[Boris Zbarsky [:bzbarsky]]

Comment on attachment 264422 [details] [diff] [review]
remove proportional units from style system

Excellent.

Comment 15

[David Baron :dbaron: (⌚️UTC-5, no longer working on Mozilla)]

Above patch checked in to trunk.