133429 - dynamically created MathML content sometimes doesn't layout as expected

Status: RESOLVED FIXED

(Core :: MathML, defect)

Description

[steve.swanson]

(See testcase, which includes instructions on how to recreate the bug.)

In certain cases, adding content to MathML documents causes existing content 
(in the run of MathML) to layout strangely. The only cases I've observed are 
with stretchy <mo>s.  (The testcase uses a big operator, but I've also seen the 
problem with fences.)

Comment 1

[steve.swanson]

Attachment: testcase

includes instructions on how to reproduce the bug

Comment 2

[steve.swanson]

Attachment: better testcase using fences

Previous testcase only failed when patch for bug 129560 was applied. This
version seems to always fail.

Comment 3

[rbs]

-> m1.0. Great testcase...

Comment 4

[rbs]

Found the problem. A state in the stretchy goes out of sync. Specifically, when
the layout of '(x)' finished, the stretchy chars for '(' and ')' recorded that 
they didn't stretch by setting their mDirection to -1. Now, when 'x' is replaced 
with a fraction, the old state causes the rendering to mess up. When the state 
is reset, the rendring works fine.

Comment 5

[rbs]

Attachment: patch to fix the bug

Comment 6

[rbs]

as usual, pinging roc & attinasi for r/sr...

Comment 7

[Marc Attinasi]

Comment on attachment 76533 [details] [diff] [review]
patch to fix the bug

sr=attinasi, as usual

Comment 8

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

Comment on attachment 76533 [details] [diff] [review]
patch to fix the bug

r=dbaron

Comment 9

[Asa Dotzler [:asa]]

Comment on attachment 76533 [details] [diff] [review]
patch to fix the bug

a=asa (on behalf of drivers) for checkin to the 1.0 trunk

Comment 10

[rbs]

Checked-in.