Issue 45049

Same problem occurs on Linux platform too. I checked on Fedora Core 3.

Kindly put in more details on the above.

Well, the problem is when we switch layout (to Bangla) and start typing 
Bangla, OO doesn't change the font to any installed Bangla font. It just shows 
boxes.

But a good news is when we manually select a Bangla font and after typing 
Bangla when we switch layout back to English, then it can detect the last used 
English font and again when we switch layout to Bangla, it can remember the 
last selected Bangla font and presents Bangla look.

The big problem is although we have selected a Bangla font, but when we type 
Danda or Double Danda, those characters comes from the default Devanagari 
font. In Windows Mangal is the default Devanagari font (also the UI font). So 
the Danda/Double Danda comes from Devanagari font. It doesn't comes from the 
font I have selected for Bangla.

So that is the main problem. I tried to explain some problems 
here:http://www.openoffice.org/issues/show_bug.cgi?id=45064

Also there is not Bangla locale in Language settings. So this is very sad.

I guess this bug is related to #53006

The problem of Bangla font characters displayed as square boxes, reported by Omi
Azad, becomes more devastating when typing some Bangla text within the "Draw"
window, as and when we need any diagram within the text. In fact the only
solution    becomes blind typing a few characters and then selecting them all
and changing them again from Tahoma to Solaiman Lipi or so, though you made it
so in the very start. 

This font issue becomes more problemsome with FC3. I don't know what happens
with FC4, have  not seen one, but, a text written with Solaiman Lipi, BN
Bidisha, or Sutonny OMJ, with text alignment "justified", becomes paras with
zagged right border in OOo 2 Beta in FC3. Though the display remains normal in
Ubuntu 504 or SuSE 92. This problem does not happen in FC3 if the font is Lohit.
Another thing happens in FC3 about inter-character spaces, specially when using
some symbols. Like a text string in Bangla, 'shabda/chhabi/dharana' becomes  a
pulp of the '/' and the Bangla characters.

Another very sad thing is faulty representation when exporting to a PDF. The
page-numbers showing in Bangla in the *.odt, *.sxw, or even *.rtf or *.doc files
become page-numbers in roman numerals in the pdf. Though i have not tried it in
Acroread or Evince. This is just Xpdf experience. 

> This font issue becomes more problemsome with FC3. I don't know what happens
> with FC4, have  not seen one, but, a text written with Solaiman Lipi, BN
> Bidisha, or Sutonny OMJ, with text alignment "justified", becomes paras with
> zagged right border in OOo 2 Beta in FC3. Though the display remains normal in
> Ubuntu 504 or SuSE 92. This problem does not happen in FC3 if the font is Lohit.

I could not make this problem. It's working fine on my Windows box. Can you
please supply some screen shorts?


>Another thing happens in FC3 about inter-character spaces, specially when using
>some symbols. Like a text string in Bangla, 'shabda/chhabi/dharana' becomes  a
>pulp of the '/' and the Bangla characters.
>


I didn't get this point. Can you please explain how can I re-make the 
problem?

Regards

-- 
Omi Azad
Contributor
Bangla Computing and Localization Projects:
Ankur: http://www.ankurbangla.org
Ekushey: http://www.ekushey.org

All these problems should be sold when Bengali is the locales, considered as CTL
(probably both already), is in VCL.xcu and in has a font in the glyph fallback
list. Then by default characters will be looked for in the present font. Only
when they are not found there, glyph fallback will be activated.

For more information: http://www.khmeros.info/tools

The danda problem probably comes from the fact that Bengali has no dandas in its
encoding, the danda code-points are shared by all scripts... and they are in the
devanagari block of characters. If glyph fallback is activated (the glyph is not
found in the present or in the CTL default font), it will automatically go to
the first font that has the code-point and that is in the list of glyph
fall-back fonts... and it will probably be a devanagari font. This is a UNICODE
problem, which has been discussed at length... but I do not think that they
concluded that dandas should be added to all scripts.