Bug 8072 – REGRESSION: text fields at connect.apple.com spill out of the containing box

• Status: RESOLVED
• Resolution: FIXED
• Component: Layout and Rendering

[Henk]

The widths of the textfields "Apple ID" and "Password" are incorrect.

The regression was intruduced starting with nightly r13567

[Darin Adler]

Actually, the widths of the text fields look fine to me (same as before). The problem seems to be the position of the text fields, not the widths.

[Darin Adler]

Created attachment 7475 [details]
reduction, showing this is some sort of min/max width issue

[Darin Adler]

I think the bug is caused by the fact that RenderTextField::calcMinMaxWidth sets the min/max width to a width based on the input element's size attribute even when the style has a width specified.

[Darin Adler]

RenderTextField::calcMinMaxWidth is clearly wrong in at least a few ways. For example, it doesn't consider border or padding. It doesn't consider the values of any CSS properties (width, min-width, max-width).

What's annoying is that we probably want most of the logic from RenderBlock::calcMinMaxWidth, but we can't use it because it calls non-virtual calcInlineMinMaxWidth or calcBlockMinMaxWidth. We can look at RenderReplaced::calcMinMaxWidth for inspiration, I guess. Or maybe Hyatt already has a good technique in mind for this.

[Maciej Stachowiak]

These are all text field regressions so they should all be P1.

[Adele Peterson]

working on this now.

[Darin Adler]

Assigning to Adele since she says she is working on this.

[Adele Peterson]

Created attachment 7503 [details]
initial patch

Dave, can you take a look at this first cut of fixing RenderTextField::calcMinMaxWidth?

[Dave Hyatt]

Comment on attachment 7503 [details]
initial patch

You shouldn't need these lines at all:

+    calcBlockMinMaxWidth();
+
+    if(m_maxWidth < m_minWidth) m_maxWidth = m_minWidth;

Other than that, I think this is right, although we probably need more tests of non-fixed widths on text fields (i.e., width:100%).