391140 - [FIXr]The second one of two input type=image 's doesn't render when they have display: table-cell

Status: RESOLVED FIXED

(Core :: Layout, defect, P2)

Description

[Martijn Wargers (dead)]

Attachment: testcase

See testcase, I just see one "Submit Query", where I would have expected 2 of them.
(although I might not expect to see that text, see bug 380144)

This regressed between 2006-01-12 and 2006-01-14:
http://bonsai.mozilla.org/cvsquery.cgi?treeid=default&module=all&branch=HEAD&branchtype=match&dir=&file=&filetype=match&who=&whotype=match&sortby=Date&hours=2&date=explicit&mindate=2006-01-12+04&maxdate=2006-01-14+08&cvsroot=%2Fcvsroot
Regression from bug 51767? Or from bug 302725?

Comment 1

[Boris Zbarsky [:bzbarsky]]

This is a regression from bug 51767.  In this case, we construct the pseudo frames for the first input.  Then we go to do that for the second input, process the table pseudo frames unconditionally (even though we didn't construct a frame for it, and it's not actually being a replaced element!) and then bail out.  ConstructFrameInternal sees that there's a new last child (the table outer from the processed pseudo frames) and thinks we actually constructed things and there's nothing more to do.

Processing pseudo frames here unconditionally is wrong.

Comment 2

[Boris Zbarsky [:bzbarsky]]

Attachment: Like so

Comment 3

[Bernd]

Comment on attachment 275561 [details] [diff] [review]
Like so

.... embarassing ....

Comment 4

[Boris Zbarsky [:bzbarsky]]

Comment on attachment 275561 [details] [diff] [review]
Like so

This is a fix to not process table pseudo frames in the edge case of an image input that is getting replaced by alt text (since that's not a replaced element).  Risk is fairly low.

Comment 5

[Boris Zbarsky [:bzbarsky]]

Fixed.