191292 - Unicode: combining macron (overbar) does not display correctly

Status: NEW

(Core :: Graphics: Text, defect)

Description

[Matthew S]

User-Agent:       Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.3a) Gecko/20021212
Build Identifier: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.3a) Gecko/20021212

The sequence ν̄ is meant to display the greek letter nu with a bar
over it.  This works in various versions of Mozilla on various versions of
Windows (i.e. Mozilla 1.1 in Windows 98), but in Linux, it is displayed as a nu
_followed_ by an overbar.

Here's a page to test it: http://gridley.acns.carleton.edu/~straitm/overbar.html

Apologies in advance if this is a subset of a larger known problem.  I'm not
very familiar with Unicode.

Reproducible: Always

Steps to Reproduce:
1. Create a page with the above character sequence in it.
2. View it.

Actual Results:  
Displays a nu followed by an overbar.

Expected Results:  
Should display a nu with an overbar over it.

Comment 1

[Boris Zbarsky [:bzbarsky]]

intl; this is a dup of the bug on combining chars being broken on Linux in
general, iirc.

Comment 2

[Simon Montagu :smontagu]

This WORKSFORME. Is it a font issue?

Comment 3

[Matthew S]

I should have been more detailed about my setup.  I'm running Redhat 8.0, kernel
 2.4.18-14 with XFree86 4.2.0 (this is what it shipped with) and the default
window manager ("Blue Curve" or whatever it's called).  The only mucking with
fonts I've done is to install some extra ones.  Here's the output of "rpm -qa |
grep font":

XFree86-font-utils-4.2.0-72
ghostscript-fonts-5.50-7
XFree86-ISO8859-2-75dpi-fonts-4.2.0-72
fonts-ISO8859-2-1.0-8
fonts-KOI8-R-100dpi-1.0-3
fontconfig-2.0-3
XFree86-truetype-fonts-4.2.0-72
urw-fonts-2.0-26
abiword-fonts-1.0.3-1
XFree86-100dpi-fonts-4.2.0-72
XFree86-base-fonts-4.2.0-72
fontconfig-devel-2.0-3
tetex-fonts-1.0.7-57
XFree86-ISO8859-15-100dpi-fonts-4.2.0-72
XFree86-ISO8859-2-100dpi-fonts-4.2.0-72
XFree86-ISO8859-9-100dpi-fonts-4.2.0-72
bitmap-fonts-cjk-0.2-2
fonts-ISO8859-2-100dpi-1.0-8
fonts-KOI8-R-1.0-3
fonts-KOI8-R-75dpi-1.0-3
chkfontpath-1.9.6-3
bitmap-fonts-0.2-2
XFree86-75dpi-fonts-4.2.0-72
XFree86-ISO8859-15-75dpi-fonts-4.2.0-72
XFree86-ISO8859-9-75dpi-fonts-4.2.0-72
fonts-ISO8859-2-75dpi-1.0-8
kon2-fonts-0.3.9b-13

Hope that some of that was useful. :)

Comment 4

[Dave Oftedal]

I have a URL that demonstrates this:

http://www.wiregauze.net/~seikai/board/viewtopic.php?t=671&mylang=en

Notice how characters that normally have macrons etc. display quite nicely.
In Unicode, however, ANYcharacter should be combineable with a combining
character. Mozilla just displayes the combining character after the char, which
is incorrect.

Comment 5

[simon annear]

Attachment: test html file

Comment 6

[simon annear]

Have created another testcase

works ok on Win XP with Moz 1.8b2
but not on Linux (Sun JDS R2) + moz 1.8b

To reproduce problem

1. open testcase (test.html) file using Moz on Linux

Expected results:
macron (overbar) is over v character as per windows

Actual results:
macron (overbar) is to right of v character

Comment 7

[srikumarks]

Attachment: Shows upper case N with a COMBINING OVERDOT in a sans-serif font.

This case shows COMBINING unicode characters are broken on MacOSX too in Firefox.

Upper case "N" when suffixed with a COMBINING OVERDOT unicode character should show a dot *above* the N. In Chrome and Safari on MacOSX, the dot appears correctly above N, but on Firefox (27.0.1) on MacOS 10.9.1, the dot appears within the top edge of the bounding box of the N letter and not above it.

Comment 8

[srikumarks]

Attachment: Shows tamil script "ri" character with a COMBINING OVERDOT

This case shows how the typesetting of COMBINING unicode characters in Firefox is broken for non-Roman scripts. 

Safari on MacOSX (10.9.1) and iOS renders this correctly, but both Chrome and Firefox overlap the COMBINING OVERDOT with the character.

Comment 9

[srikumarks]

Here is an additional example to show that COMBINING unicode characters don't render correctly -

vīṇa

The "i" character in "vina" has a COMBINING MACRON following it. When used with "i" in a sans-serif font, the dot above the "i" should be replaced with the macron (which happens correctly in Chrome and Safari on MacOS 10.9.1), but in Firefox, the macron is placed slightly to the left of the "i" (in Firefox 27.0.1) and the dot over the "i" is not removed.

Comment 10

[Jonathan Kew [:jfkthame]]

This is dependent on the specific font being used. In my testing on OS X, your "vīṇa" example looks fine in many common fonts I tried, though some (such as Georgia and Verdana) fail to handle it.

Comment 11

[srikumarks]

(In reply to Jonathan Kew (:jfkthame) from comment #10)
> This is dependent on the specific font being used. In my testing on OS X,
> your "vīṇa" example looks fine in many common fonts I tried, though some
> (such as Georgia and Verdana) fail to handle it.

Right. When the same text is viewed in Chrome and Safari (using Georgia, say), what happens is that the "v" and "a" are in Georgia, but the "i" (with macron above) and "n" (with dot below) seems to fall back to Times. I guess this because when I copy the text from these browsers and paste into TextEdit, those are the fonts that show up. 

Strangely, the same thing shows up when I copy the text from Firefox too, so I can't rule out whether that fallback is happening in TextEdit instead of in the browsers.

The Firefox "font" panel does say the font is Georgia throughout the word.

In summary, whatever fallback Chrome and Safari seem to be doing is working better than the fallback choices in Firefox.

Comment 12

[srikumarks]

For the tamil "ri" example, the fonts that show up in TextEdit when I force the font-family to "Times" are "Times" for the "ri" character itself and "Lucida Grande" for the COMBINING OVERDOT. So it looks like the fallback choices are what need to be tweaked.