OpenOffice Language Problem

[Jmadero]

Binary package hint: openoffice.org

Hi All,

This problem is in Hardy Heron.

I am trying to get a particular language working correctly in OpenOffice (it works great in pidgin).

I think the easiest way to replicate would be to do the following:

Install Telugu language (system, administration, language support), you'll also need to install the complex characters.

Now type in pidgin the letters K then e, this will make a letter with a subletter on top which is pronounced "kaa"

If you try the same thing in OO you'll get the two letters separate. The only way I have figured out how to combine the two is to do "e" "k" "e" which gives the subletter at the beginning than the correct "kaa" after it.

Also, I can't adjust the size of the font at all when I switch to Telugu.

Maybe I'm just doing something wrong but I've tried quite a few different things and can't get it working. Thanks to everyone working on these projects.

[Chris Cheney (ccheney)]

Can you look into this issue for me?

Thanks!

Chris

[Jmadero (jmadero)]

How do you mean look into the issue? I have verified the problem multiple times, I will test it with Intrepid once it's released.

[Arne Goetje (arnegoetje)]

Jmadero wrote:
> How do you mean look into the issue? I have verified the problem
> multiple times, I will test it with Intrepid once it's released.
>

He asked me to look into this issue, since I'm the i18n guru in
Canonical. ;) I will do so, once the urgent Intrepid issues have been
resolved.

Could you please prepare a test case for me:

 * which locale settings are you using?
 * the output of 'im-switch -l'
 * I suppose you use SCIM with the phonetic Telugu IME? Or something else?
 * a string to type (keyboard sequences -- which keys do I have to press?)
 * expected output (screenshot)
 * screenshot of the faulty output
 * any other applications besides openoffice.org affected?

I will try to figure out where exactly the problem lies.

Thanks.

[Jmadero (jmadero)]

 * which locale settings are you using?
Not sure exactly what this is but I believe it is Pacific, USA, US English Keyboard is default

 * the output of 'im-switch -l'

Your input method setup under en_US locale as below.
=======================================================
No private configuration can be defined for root account.
=======================================================
The system wide default is pointed by "/etc/alternatives/xinput-all_ALL" .
xinput-all_ALL - status is auto.
 link currently points to default
default - priority 10
default-xim - priority 0
none - priority 0
Current `best' version is default.
=======================================================
The available input method configuration files are:
default default-xim none scim scim-bridge scim-immodule th-xim

 * I suppose you use SCIM with the phonetic Telugu IME? Or something else?
Yes I am using SCIM with Telugu installed from the repositories. I use English by default and switch to Telugu with Cntrl+space

 * a string to type (keyboard sequences -- which keys do I have to press?)
Not sure how many keys you'd like to test but there are several problems so I'll do a few:

1. Just do the key "k" in Telugu script in OO then in Text Editor (I am using Gnome with Ubuntu)

2. Sub letters
Do this string: kdk

3. Top letters (not sure what they are called)
Do kf

It looks like all subletters don't work and that the top letters have problems as well.

 * expected output (screenshot)
Attached is the output from text editor that is correct (it works in firefox as well, not sure what other programs I can test in)

 * screenshot of the faulty output
OO 2.4.1 (have tried in earlier versions as well)

 * any other applications besides openoffice.org affected?
None that I know of

[Jmadero (jmadero)]

 * which locale settings are you using?
Not sure exactly what this is but I believe it is Pacific, USA, US English Keyboard is default

 * the output of 'im-switch -l'

Your input method setup under en_US locale as below.
=======================================================
No private configuration can be defined for root account.
=======================================================
The system wide default is pointed by "/etc/alternatives/xinput-all_ALL" .
xinput-all_ALL - status is auto.
 link currently points to default
default - priority 10
default-xim - priority 0
none - priority 0
Current `best' version is default.
=======================================================
The available input method configuration files are:
default default-xim none scim scim-bridge scim-immodule th-xim

 * I suppose you use SCIM with the phonetic Telugu IME? Or something else?
Yes I am using SCIM with Telugu installed from the repositories. I use English by default and switch to Telugu with Cntrl+space

 * a string to type (keyboard sequences -- which keys do I have to press?)
Not sure how many keys you'd like to test but there are several problems so I'll do a few:

1. Just do the key "k" in Telugu script in OO then in Text Editor (I am using Gnome with Ubuntu)

2. Sub letters
Do this string: kdk

3. Top letters (not sure what they are called)
Do kf

It looks like all subletters don't work and that the top letters have problems as well.

 * expected output (screenshot)
Attached is the output from text editor that is correct (it works in firefox as well, not sure what other programs I can test in)

 * screenshot of the faulty output
OO 2.4.1 (have tried in earlier versions as well)

 * any other applications besides openoffice.org affected?
None that I know of

[Jmadero (jmadero)]

Sorry I double posted and only got one of the two attachments, here is the second attachment:

[jscc88 (jscc88-deactivatedaccount)]

this problem dont be in the 3.0 version. this is not a ubuntu bug

[Jmadero (jmadero)]

I sitll have this problem in OOo3 (completely fresh install of Ubuntu Intrepid with OOo3 installed). Did a test run this morning and have the exact same problem...if it's not a Ubuntu problem where should I post it?

[Julian Alarcon (julian-alarcon)]

The user Juan Sebastian Cobaleda Cano is a troll, or something. We, in the Ubuntu-Co Team are checking all his changes in Launchpad. Sorry for the problems.

[Chris Cheney (ccheney)]

Arne,

Pinging you about this bug, it appears the user attached the relevant information.

Thanks,

Chris

[Arne Goetje (arnegoetje)]

OK, this seems to be an upstream bug.

I did some triaging and this is what I found:

I tested on Hardy and Jaunty (oo.o 2.4.1 and oo.o 3.0 respectively):
 * start oowriter in en_US locale
 * Initial font was set to "Nimbus Sans" in 2.4.1 and "Liberation Sans" in oo.o 3.0
 * in a new document type 'kdk'
 * the font changed to "DejaVu Sans"
 * rendering appears to be broken.
 * I selected the text and selected "Lohit Telugu" as font.
 * rendering is correct.
 * I deleted the whole text
 * font switched back to "Nimbus Sans" or "Liberation Sans" respectively
 * typed 'kdk' again
 * this time oo.o remembered that I chose "Lohit Telugu" before, selected that font and rendered the text correctly.

 * start oowriter in te_IN locale
 * interface is in Telugu and default font is "Lohit Telugu"
 * typed 'kdk'
 * rendering is correct.

To test if this is a fontconfig problem, I ran the following test:
 * I ran 'fc-match -s' with the following arguments each:
  * sans (lists preferred fonts for sans)
  * LiberationSans
  * Times (the default font setting in oo.o)
  * Helvetica (default font for sans in oo.o IIRC)
 In all outputs, several fonts were listed before "Lohit Telugu" showed up in the list. I checked those fonts with fontforge and none of them contained Telugu characters. So, when using fontconfig, it should have fallen through all these fonts until hitting "Lohit Telugu", why it stopped on "DejaVu Sans", I don't know.

 * fc-match -s :lang=te
  prefers "Lohit Telugu" in front of all other fonts
 * LANG=te_IN fc-match -s sans
  also prefers "Lohit Telugu" over all other fonts.

Interestingly 'fc-match -s Helvetica' even lists 'Nimbus Sans L' in front of 'Liberation Sans' on Jaunty. So, I really don't know why oo.o defaults to "Liberation Sans" by default. This suggests that it's not a fontconfig problem, since fontconfig gives back the correct preference list when queried.

(On a side note: also gnome-terminal gives a wrong rendering for 'kdk', not as bad as oo.o, but still wrong. Seems to be a problem with some not applied contextual shaping rules... but that would be a separate bug)

[Jmadero (jmadero)]

That's really interesting. I hadn't realized the fonts were changing like that. You're right that when I switch to Telugu in SCIM inside of OOo the first time I push k it goes to DejaVu Sans, then if I change it back to Lohit Telugu it works and the font is rendered (mostly) correctly.

Thanks for the update, I'm glad that it seems like it's being pinpointed. I might try the te_IN locale and see what I think. I'm working on learning the language and having it on the computer is a huge help.

As for the terminal, should I make a new bug with this problem or will someone else handle that?

Thanks again for the hard work, we all appreciate it.

[Chris Cheney (ccheney)]

Does this bug sound the same/similar as bug 16374 (and the upstream bug?), or is this different?

[Arne Goetje (arnegoetje)]

Chris Cheney wrote:
> Does this bug sound the same/similar as bug 16374 (and the upstream
> bug?), or is this different?

Sounds related. Also maybe bug 273641 (waiting for a test case to confirm).