Bug 146181 - Incorectly displayed ť, ľ in SK locales
Summary: Incorectly displayed ť, ľ in SK locales
Status: RESOLVED FIXED
Alias: None
Product: SUSE Linux 10.1
Classification: openSUSE
Component: OpenOffice.org (show other bugs)
Version: Beta 2
Hardware: Other Other
: P5 - None : Critical (vote)
Target Milestone: ---
Assignee: Jan Holesovsky
QA Contact: E-mail List
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2006-01-27 16:02 UTC by Ladislav Michnovic
Modified: 2006-02-13 15:59 UTC (History)
4 users (show)

See Also:
Found By: Other
Services Priority:
Business Priority:
Blocker: ---
Marketing QA Status: ---
IT Deployment: ---


Attachments
A screenshot when the fonts are correct. (49.42 KB, image/png)
2006-01-27 18:16 UTC, Petr Mladek
Details
A screenshot when the fonts are broken. (45.51 KB, image/png)
2006-01-27 18:16 UTC, Petr Mladek
Details
"good" cache (121.31 KB, application/x-bzip2)
2006-01-30 11:13 UTC, Petr Mladek
Details
"wrong" cache (156.68 KB, application/x-bzip2)
2006-01-30 11:26 UTC, Petr Mladek
Details

Note You need to log in before you can comment on or make changes to this bug.
Description Ladislav Michnovic 2006-01-27 16:02:23 UTC
I see in menu squares instaed of ť or ľ. Others letters like č,š are displayed correctly. My locale is sk_SK.UTF-8 .
Comment 1 Petr Mladek 2006-01-27 18:13:15 UTC
I seems that a problem with font cache is here again. I have started OOo in windowmaker (where the original problem appeared) in sk_SK locales several times. It sometimes started correctly and sometimes incorrectly.

Kendy, I think that you have already solved a similar problem half an year ago. At the time, it was deterministic. It started well and wrong in turns. It seems to be more random now. However, I started it sometimes in sk_SK locales and sometimes sk_SK.UTF-8 locales. So, there could be a system dependant on the locales.

Also note, that some problems with font cache have been discussed on research@suse.de in the thread Firefox-1.5 is slower last week.

Last note. I reproduced it on beta2. I did not see such problems on beta1.
Comment 2 Petr Mladek 2006-01-27 18:16:09 UTC
Created attachment 65479 [details]
A screenshot when the fonts are correct.
Comment 3 Petr Mladek 2006-01-27 18:16:59 UTC
Created attachment 65481 [details]
A screenshot when the fonts are broken.
Comment 4 Petr Mladek 2006-01-27 19:01:37 UTC
Ah, I found a system. OOo starts with correct fonts after I remove the font cache ~/.fonts.cache-2 or after the cache is updated by a "smart" application, such as gimp.

So, it is most probably related to the new fontconfig as discussed on research@suse.de, http://mailman.suse.de/mlarch/SuSE/research/2006/research.2006.01/msg00266.html

It seems to be a beta2-specific problem.

It seems to be quite common problem, which must be fixed before we release 10.1, so I'll increase severity to crytical.

Mike, do you have any idea what goes wrong?
Comment 5 Mike Fabian 2006-01-28 09:57:00 UTC
No, I don't know what is going wrong. But I know that there are
many remaining bugs in fontconfig and that it usually
helps to delete all caches in /var/cache/fontconfig and ~/.fonts.cache-2.
Of course that's a workaround only.

I am very interested in broken ~/.fonts.cache-2 files, i.e. if
you have a system where a problem occurs which can be fixed
by moving the ~/.fonts.cache-2 out of the way, then please don't
delete that ~/.fonts.cache-2 file, just move it somewhere else
and send it to me. That would help me to investigate what is going
wrong.

I fixed a problem with ~/.fonts.cache-2 Friday evening, i.e. you
can also try the fontconfig which I submitted Friday evening
(still in /work/src/done/STABLE/fontconfig, mbuild job magellan-mfabian-80)
and see whether that helps.
Comment 6 Petr Mladek 2006-01-30 11:13:43 UTC
Created attachment 65603 [details]
"good" cache

This cache has been generated when the gimp was started. A similar (such a big cache) was generated also log in. When such a big chache is present, OOo starts correctly, witch correct font.
Comment 7 Petr Mladek 2006-01-30 11:26:00 UTC
Created attachment 65605 [details]
"wrong" cache

This cache was generated when OOo was started. OOo starts with wrong/broken font when such a small cache is present on the system.

To avoid misunderstanding. The most important thing is what cache is present before OOo is being started. When the big cache is available, it starts well. When the small cache is available, it starts with wrong fonts.

By another words. When OOo is beeing started and the good cache is present, the cache is regenerated to the small variant, OOo starts with good fonts but the cache is broken and OOo starts with wrong fonts next time.

When OOo is beeing started with the wrong cache, the cache is not regenerated, OOo starts with wrong fonts, the cache is still broken and OOo starts with wrong fonts next time as well.
Comment 8 Petr Mladek 2006-01-30 11:53:02 UTC
Wow, I updated fontconfig from STABLE and it works now. OOo still produces small cache, gimp big cache, OOo works everytime, though.

So, it works and we would close this bug. However, I would like to be sure that there is not a problem with the cache size.

Is it correct that gimp produce a bigger font cache than OOo?

I think that either gimp adds its own private fonts or OOo disables some system fonts. There are two problems with it:

1. It would be ugly if OOo does not find some system fonts.
2. It takes quite a long time to regenerate the user font cache. The applications start slowly then. So, we should avoid the situation when some 
applications modify the cache this way. We should probably open another bug for this.

Mike, what do you think about it?
Comment 9 Mike Fabian 2006-01-30 12:10:25 UTC
Petr> Is it correct that gimp produce a bigger font cache than OOo?

No, that cannot be correct.
Comment 10 Mike Fabian 2006-01-30 12:13:57 UTC
Petr> I think that either gimp adds its own private fonts or OOo disables
Petr> some system fonts.

I don't think this is what is happening.

All applications using fontconfig should see the same set of fonts.
If not something is badly wrong.

Comment 11 Ladislav Michnovic 2006-02-10 21:01:33 UTC
I cannot reproduce this in Beta 3. 
(But I cannot be 100% sure.)
Comment 12 Mike Fabian 2006-02-10 23:35:05 UTC
The fontconfig packages in Beta3 are unfortunately very broken.

Please try the updated packages available from

    ftp://ftp.suse.com/pub/projects/m17n/10.1/RPMS

Please update both the "fontconfig" and the "fonts-config" packages!

As I believe this problem has been fixed, I'm closing this bug as FIXED.

Please REOPEN if the problem remains with the updated packages from
the above ftp URL.
Comment 13 Ladislav Michnovic 2006-02-13 12:23:39 UTC
I'm using beta3. 
After I have updated fontconfig*, groff and fonts-config everything related with X begun crashing. Applications, also gdm. I tried to launch yast2. I can attach strace and core if needed. Actually I had no problem in beta 3 with original fontconfig*, fonts-config, groff. So I had to downgrade.
Comment 14 Mike Fabian 2006-02-13 14:25:05 UTC
You did call "SuSEconfig --module fonts"?
Comment 15 Ladislav Michnovic 2006-02-13 14:34:25 UTC
No. 
Comment 16 Mike Fabian 2006-02-13 14:40:06 UTC
Does it help if you do it now?

If not, does "fc-cache -v -f" (called as root) help?
Comment 17 Ladislav Michnovic 2006-02-13 15:11:26 UTC
Runing 
 SuSEconfig --module fonts
did not helped. Try to run Yast2 ends with segfault.
But 
 fc-cache -v -f
helped. Yast2 runs fine.
 Probably this information could be mentioned on the page
http://en.opensuse.org/Bugs:Most_Annoying_Bugs .
Comment 18 Ladislav Michnovic 2006-02-13 15:13:35 UTC
Characters in OOo are displayed correctly after update.
Comment 19 Mike Fabian 2006-02-13 15:24:40 UTC
Ladislav>  Probably this information could be mentioned on the page
Ladislav> http://en.opensuse.org/Bugs:Most_Annoying_Bugs .

Done.

Comment 20 Mike Fabian 2006-02-13 15:26:38 UTC
Ladislav> Characters in OOo are displayed correctly after update.

Can you give some more information please?
Comment 21 Ladislav Michnovic 2006-02-13 15:35:35 UTC
I can not reproduce this bug [description] any more. So this bug can be closed.
Comment 22 Mike Fabian 2006-02-13 15:59:08 UTC
Closing as FIXED according to comment #21.