413928 - Hebrew numbering letters displayed in reverse

Status: VERIFIED FIXED

(Core :: Layout: Text and Fonts, defect, P2)

Description

[Ori Avtalion (salty-horse)]

Attachment: test-case

Using nightly build:

The hebrew numbering displayed with list-style-type:hebrew have the letters in reverse (left-to-right).
See the attached test-case for an example.


Using Firefox 2.0, the letters appear in the correct right-to-left order.

Also note that in 2.0, a different bug is visible:
The dot between the number and the text is in the wrong side: ".<number> <text>".
It is already fixed in trunk.

Comment 3

[Simon Montagu :smontagu]

I'm already working on this. I have a patch for the bug as reported, but it
regresses the issue mentioned at the end of comment #0:

> Also note that in 2.0, a different bug is visible:
> The dot between the number and the text is in the wrong side: ".<number>
> <text>".
> It is already fixed in trunk.

Comment 4

[Simon Montagu :smontagu]

Attachment: Corrected test-case

I see wrong rendering in Safari (Windows and Mac) in 0 through 4 in the LRT section, but that's because of a missing line in the testcase. This version displays as expected, except that in the RTL section the negative numbers are displayed -3, -2, -1 in the list numbering and 3-, 2-, 1- in the list items (We do the same).

Comment 6

[Uri Bernstein]

Regressed by bug 370588, according to the regression date.

Comment 7

[Simon Montagu :smontagu]

Attachment: Patch

I really have no choice here but to remove the codepaths for IBMBIDI not defined and for systems with no bidi support, since I have no way to test them.

Comment 8

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

Why are you removing the hebrewZero support?

Why does mSuffix need to be stored in the nsBulletFrame?

Comment 9

[Simon Montagu :smontagu]

Attachment: Bigger testcase adjusted for changes in the algorithm

I made a few changes in the algorithm that are reflected in this new testcase. We now use U+05F3 HEBREW PUNCTUATION GERESH to mark numbers >= 1000 and bail to decimal for numbers >= 1,000,000 or < 1 (not 0 as before). Except for 0, this makes us interoperable with Safari.

Comment 10

[Simon Montagu :smontagu]

(In reply to comment #8)
> Why are you removing the hebrewZero support?

The Hebrew numbering system, like the Roman, is old enough that it has no concept of zero, and using the modern word for "Zero" mixed in with the numbers has never made sense to me. We have to bail to decimal for negative numbers anyway.

> Why does mSuffix need to be stored in the nsBulletFrame?

That's preparation for fixing the //XXX comment:
  // XXX For some of these systems, "." is wrong!  This should really be
  // pushed down into the individual cases!
I can cut it for now, if you prefer.

Comment 11

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

Comment on attachment 300741 [details] [diff] [review]
Patch

Please cut mSuffix for now, yes. r+sr with that

Comment 12

[Simon Montagu :smontagu]

Checked in, with reftests.

Comment 13

[Al Billings [:abillings - ex-MoCo]]

Verified in Mozilla/5.0 (Macintosh; U; Intel Mac OS X 10.5; en-US; rv:1.9pre) Gecko/2008050704 Minefield/3.0pre.