56168 - Missing character after closing inline tag which spans lines

Status: VERIFIED FIXED

(Core :: DOM: Editor, defect, P3)

Description

[Simon Fraser [no longer active]]

I see a case where a character typed after closing an inline tag disappears. To 
repro:

1. type abc
2. Hit cntrl-B for bold
3. type 123
4. Hit return for a new line
5. type 456
6. Hit cntrl-B to turn off bold
7. Type a single char, e.g. 'a'.
   Note that the char does not appear, and the caret goes into an odd place
8. Type another char. That one will  show up.

This only seems to happen if the style spans lines.

Comment 1

[Simon Fraser [no longer active]]

Rtm, this makes typing styled text hard.

Comment 2

[Simon Fraser [no longer active]]

Seen on the branch, debug and opt builds.

Comment 3

[Joe Francis]

i see it too; accepting bug; investigating...

Comment 4

[Simon Fraser [no longer active]]

M18

Comment 5

[Joe Francis]

Attachment: diffs of editor/base/nsHTMLEditRules.cpp

Comment 6

[Joe Francis]

simon, kin: ready to walk you through the review.  do we want this for rtm?
fairly low risk but not totally trivial.  it is on important html typing path,
but will only be executed while removing styles while typing.

Comment 7

[Joe Francis]

attaching a new diff that is appropriately more paranoid about null values

Comment 8

[Joe Francis]

Attachment: new diff of editor/base/nsHTMLEditRules.cpp

Comment 9

[Kathleen :Brade]

add rtm need info on status whiteboard to get it in some bugzilla queries

Comment 10

[Phil Peterson]

PDT marking [rtm-] because (1) it's somewhat of an edge case that the inline
style would span <br>s (2) you can easily reposition the insertion point with
the mouse when that one character moves the insertion point to a bad place.

Comment 11

[Simon Fraser [no longer active]]

This isn't just a case of the insertion point moving to the wrong place, it's 
data loss (albeit of 1 char). The first character I type after closing the style 
goes into the abyss.

However, given the size of the patch, I agree that some caution is justified.

Comment 12

[Kathleen :Brade]

requesting re-evaluation of this bug; I'm a touch typist and don't notice that 
Composer ate some of the letters/characters I type.

Due to the size of the change/diff, maybe PDT would be more re-assured with more 
reviewers than the standard 2?  

Joe--can you check this in on the trunk so it could be tested there?

Comment 13

[Peter Trudelle]

PDT says rtm-, patch looks too large for marginal benefit.

Comment 14

[Joe Francis]

well, i was requested not to check in on the trunk for the mean time, in order to 

better keep the trunk and branch in sync...

Comment 15

[rubydoo123]

moving to future since pdt deems it not rtm critical

Comment 16

[Joe Francis]

moz 0.9

Comment 17

[Joe Francis]

this will definitely land in time for 0.8

Comment 18

[Joe Francis]

fixed

Comment 19

[sujay]

verified in 3/22 build.