421955 - combining mark at start of text run rendered as NUL hex box or garbage

Status: RESOLVED FIXED

(Core :: Graphics, defect, P2)

Description

[Karl Tomlinson (:karlt)]

Attachment: testcase

Comment 1

[Karl Tomlinson (:karlt)]

Attachment: use a source consistent with sourceOffset

Comment 2

[distler]

See also this testcase

  https://bugzilla.mozilla.org/attachment.cgi?id=301824

(for bug 416065)

Comment 3

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

Comment on attachment 308470 [details] [diff] [review]
use a source consistent with sourceOffset

oops

Comment 4

[Damon Sicore (:damons)]

Comment on attachment 308470 [details] [diff] [review]
use a source consistent with sourceOffset

a1.9+=damons

Comment 5

[Karl Tomlinson (:karlt)]

Checked in:

http://bonsai.mozilla.org/cvsquery.cgi?module=PhoenixTinderbox&branch=HEAD&cvsroot=%2Fcvsroot&date=explicit&mindate=1205218994&maxdate=1205219043&who=karlt%2B%25karlt.net

Comment 6

[Jesse Ruderman]

What's the best way to test this?  Can the testcase be compared to something else, or should we just test that it renders the same way each time?

== 421955-1.html 421955-1.html

Comment 7

[Karl Tomlinson (:karlt)]

I don't think testing for rendering the same way each time is enough for the test case here.

I was wondering about comparing testcases each with two words, the second of which begins with a combining mark.  One of the testcases would need a (CSS?) trick to make the second word begin a new text run.

Perhaps the trick could be to specify a slightly different font-family list where the first family is a garbage family name that doesn't exist.  Any better ideas?

Comment 8

[Jesse Ruderman]

Why not?  Before the patch, it wasn't rendering the same way each time (wrt reloads).

Comment 9

[Jesse Ruderman]

Fwiw, bug 421234 was also fixed by this patch and should be easy to reftest.

Comment 10

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

That would work. But why couldn't you put the second word in its own DIV to get the second textrun?

Comment 11

[Karl Tomlinson (:karlt)]

(In reply to comment #10)
> But why couldn't you put the second word in its own DIV to get
> the second textrun?

Maybe, if that could still render the same as without the DIV (when the mark is not at the beginning of the text run).

(In reply to comment #7)
> I was wondering about comparing testcases each with two words, the second of
> which begins with a combining mark.  One of the testcases would need a (CSS?)
> trick to make the second word begin a new text run.

But I'm starting to think that this is not a good idea.
There is a good chance that the mark combining with the space (in the single text run case) could change the width of the space.

(In reply to comment #8)
> Why not?  Before the patch, it wasn't rendering the same way each time (wrt
> reloads).

You are right.
"== 421955-1.html 421955-1.html" would provide some level of testing.

Comment 12

[Karl Tomlinson (:karlt)]

Checked in a reftest comparing rendering of U+0305 COMBINING OVERLINE with and without other characters appended.

http://bonsai.mozilla.org/cvsquery.cgi?module=PhoenixTinderbox&branch=HEAD&cvsroot=%2Fcvsroot&date=explicit&mindate=1206494940&maxdate=1206495002&who=karlt%2B%25karlt.net
http://bonsai.mozilla.org/cvsquery.cgi?module=PhoenixTinderbox&branch=HEAD&cvsroot=%2Fcvsroot&date=explicit&mindate=1206499860&maxdate=1206499861&who=karlt%2B%25karlt.net