109461 - Some Unicode characters incorrectly rendered in OSX

Status: VERIFIED DUPLICATE of bug 111728

(Core :: Internationalization, defect, P4)

Description

[PTRourke]

From Bugzilla Helper:
User-Agent: Mozilla/5.0 (Macintosh; U; PPC Mac OS X; en-US; rv:0.9.5+)
Gecko/20011109
BuildID:    2001110908

Certain Unicode characters are being incorrectly rendered in OS X.  For
instance, the letter pi (u+03c0) is rendered differently by different fonts,
sometimes as a permille sign, sometimes as a Latin capital I with acute accent.
 A PNG screen shot will be attached showing correct rendering in OmniWeb and
incorrect rendering in Mozilla.

The rendering is usually correct on Windows (haven't checked it in a couple of days)

Reproducible: Always
Steps to Reproduce:
1. Install a Unicode font with support for basic and extended Greek on OS X
2. Open the given URI in both OmniWeb and Mozilla
3. Note the differences.

Actual Results:  u+03c0 always rendered incorrectly, some other glyphs rendered
incorrectly in certain circumstances.  

Expected Results:  u+03c0 rendered as Greek letter pi, lower case.

Followup to fixed bug 90804.

Comment 1

[PTRourke]

Attachment: Screenshot showing rendering problems in Greek, and comparing correct rendering in OmniWeb

Comment 2

[Roy Yokoyama]

teruko: please confirm.

cc'ing shanjian for font related issues

Comment 3

[nhottanscp]

Reassign to ftang.

Comment 4

[Frank Tang]

>1. Install a Unicode font with support for basic and extended Greek on OS X
Where can we get such font ?
>2. Open the given URI in both OmniWeb and Mozilla

Which URL ? I don't see any greek at http://www.methymna.com/
Also, I cannot view the attached png, can you ?

Comment 5

[Frank Tang]

give to nhotta

Comment 6

[PTRourke]

Changed URL to more recent, more accessible version of problem page.  Fonts 
listed in bug 90804. This is a follow-up to bug 90804, and the problem can be 
seen in the URL on that page as well.  

Main problem I've noticed is that the letter lowercase pi is rendered 
incorrectly; in a new version of Athena Unicode, it is rendered as the permille 
sign; in Cardo, it is rendered as uppercase I with acute accent above (shown in 
screenshot, which I can read from this, another computer (Windows)).  Windows 
Mozilla renders correctly.

Also, uppercase alpha with smooth aspirate and acute accent is not always 
rendered, nor is terminal sigma: they are rendered properly in the lower text, 
but not in the heading.

Comment 7

[PTRourke]

Note, by the way, that the page listed has been tested and found to work in the 
following browsers, with the compatible fonts listed in previously cited bug:

Mozilla for Windows (several versions), Linux (RedHat 6.2, 7.0);
OmniWeb 4 (OSX);
Netscape 6 for Windows (all three versions);
Netscape 4.7 for Windows (several versions);
IE 4, 5, 6;
Opera beta 6.0 for Windows (several versions);
NetPositive for BeOS 5.

Improper rendering is peculiar to Mozilla/OS X (of browsers which support 
Unicode). Additional screenshots can be provided as gifs or pngs.

Comment 8

[PTRourke]

Attachment: Gif version of attachment 1 (converted with Adobe 5.5 for Windows)

Comment 9

[Greg K.]

There have been other bugs like this. Let's get a complete Unicode test suite
together and validate Mozilla against it to avoid little bugs like this one.

Let's do a proper job of implementing Unicode on Mozilla.

Comment 10

[PTRourke]

Only problem is, you'd need quite a group of people to complete coverage. I can
put together test documents for ancient Greek (that's my schtick; I've been
keeping close watch over browser support for ancient Greek for Unicode in all
the major browsers - by which definition Amaya counts as major), but you'll need
to find people who can handle all the other scripts.

Anyway, Mozilla for Windows does not have a problem with my particular bug, so
I'm guessing that it's  something peculiar to Mozilla's implementation for OS X
(done in response to the above cited bug). 



 

Comment 11

[Henri Sivonen (:hsivonen) (not reading bugmail until 2025-02-20)]

Related bugs: bug 111728 and bug 111731.

Comment 12

[Greg K.]

PTRourke, can you still reproduce this problem under 0.9.6?

Comment 13

[PTRourke]

I'm seeing it in 2001112705.  For instance, in the Cardo font, the pi is
rendered as an I with an acute accent; in the Athena Unicode font (a version
that I have which is more recent than the publicly available version), it is a
permille.  The Code2000 font is a complete mess: less than half of the glyphs
are correct, and there are major line spacing problems. 

I'll provide a test page and screen shots in each font, as well as links to each
font, except the Athena font, which I can perhaps provide off-bugzilla (it's not
currently being distributed, but will be).

Code2000: http://home.att.net/~jameskass/
Cardo: http://members.telocity.com/~perryd/cardofnt.html

 

Comment 14

[PTRourke]

Attachment: Text of passage from Euripides, poem of Sappho

This is the file used for the tests attached from here.

Comment 15

[PTRourke]

Attachment: Screenshot of test page (59467) as displayed in 2001112705 in OS X with Athena Unicode font

Comment 16

[PTRourke]

Attachment: Screenshot of test page (59467) as displayed in 2001112705 in OS X with Cardo font

Comment 17

[PTRourke]

Attachment: Screenshot of test page (59467) as displayed in 2001112705 in OS X with Code2000 font

Comment 18

[PTRourke]

Attachment: Screenshot of test page (59467) as displayed in OmniWeb in OS X with Cardo Font

This is the preferred rendering for this text.

Comment 19

[PTRourke]

Attachment: Screenshot of test page (59467) as displayed in OmniWeb in OS X

This is the preferred rending of this text.

Comment 20

[PTRourke]

Note: above attachment (that in comment 19) is screenshot with Code2000 font.

Comment 21

[PTRourke]

Attachment: Screenshot of test page (59467) as displayed in 2001111903 in Windows XP Pro with Cardo font

This is the ideal rendering.

Comment 22

[Frank Tang]

Ok, I see two problems here
1. some Greek charcters does not display in the first line but display in the
following lines. (the first and the last character)- "is that a hint for debugging?)
2. We does not use the preferred font to display them. It seems we use Japanese
font instead. 

P4.

Comment 23

[PTRourke]

Yes, those problems are present as well. But note that when I changed the style 
attribute to indicate different fonts (to create the different screenshots, I 
merely changed the name of the preferred font), I got different renderings for 
the pi (u+03c0) code point, none of them correct, while I did get the correct 
rendering in another OS X browser (and of course in Mozilla in Windows).  And I 
got nearly readable renderings for two of the fonts, while the third was 
completely worthless.  So you are doing *something* with that font-family 
preference.

In the Edit > Preferences >> Fonts I'm using Cardo as my Unicode font.

Comment 24

[nhottanscp]

I am not sure if Frank wants to do this after he comes back.
Accept for now and set 0.9.9.

Comment 25

[PTRourke]

Anything going on with this bug? It makes Mozilla unusable for reading Unicode
Greek on OS X, and probably for other purposes that require, for instance, the
Greek letter *pi* (as in the letter used to represent the ratio of circumference
to diameter . . .).

Comment 26

[nhottanscp]

Reassign to ftang. He is changing Mac font code for other bug.

Comment 27

[Frank Tang]

accept 

*** This bug has been marked as a duplicate of 111728 ***

Comment 28

[Teruko Kobayashi]

Verified as dup.