373295 - min width computation broken for text split between two text nodes when first is continuation

Status: RESOLVED FIXED

(Core :: Layout: Text and Fonts, defect)

Description

[Jesse Ruderman]

I found this bug by making dynamic changes to layout/reftests/bugs/351641-1a.html.

Comment 1

[Jesse Ruderman]

Attachment: testcase (text should not overlap border; should match reference)

Comment 2

[Jesse Ruderman]

Attachment: reference

Comment 3

[Jesse Ruderman]

Regressed between 2006-12-07 and 2006-12-08, perhaps due to the reflow branch landing.

Comment 4

[Bernd]

looks like the minwidth change does not get propagated

Comment 5

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

No, it looks like we have problems computing min width for an unwrappable piece of text split between two separate text frames.

Comment 6

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

From stepping through in the debugger, CanBreakBetween is returning true, which is a lie.

Comment 7

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

...which is because AddInlineMinWidth is designed to process all continuations at once, but CanBreakBetween looks at the offset/length of the particular continuation passed in (which in AddInlineMinWidth is always the first).  So it's determining that yes, we can break between "abc " and "ghi", but that's not what we want to be asking.

Comment 8

[Robert O'Callahan (:roc) (email my personal email if necessary)]

Works for me in a new-textframe build. It doesn't block new-textframe, but depends on it...

Comment 9

[Mike Connor [:mconnor]]

punting remaining a6 bugs to b1, all of these shipped in a5, so we're at least no worse off by doing so.

Comment 10

[Eli Friedman]

WORKSFORME?

Comment 11

[Robert O'Callahan (:roc) (email my personal email if necessary)]

FIXED by new textframe landing, actually.

Comment 12

[Jesse Ruderman]

reftest checked in.