[kde-russian] Re[2]: Tibetan input

[kde-russian] Re[2]: Tibetan input

[Grisha Mokhin]

Hello Juraj,

JB> we've been consulting these things with Eric Bischoff. Our best
JB> bet is to try to summarize, what Tibetan characters are NOT in
JB> Unicode and try to file a proposal to add them there. Unicode
JB> standard is constantly improving and there should be no problem
JB> with this.

JB>  Would you like to help this?
Yes, definitely. Actually, from lgm.ttf font I can see for sure that
255 symbols are quite enough for all Tibetan input, including
punctuation and complex Sanskrit stacks for mantras, that are very
common in Tibetan texts. However, most of Tibetan letters are combined
from top to bottom, and sometimes there are four symbols combined to
give a single Tibetan letter, like HUM.

The existing Tibetan Unicode page was not designed with this feature
in mind, so it is not applicable for real life, and we should modify
it or have another codepage assigned.

We can even try to make an 8 bit codepage for Tibetan, because with
smart design all basic Tibetan characters can be fit into 128 symbols.

JB> Second thing: as I don't have Windows license, I can't try your
JB> program. Could you please explain how are the characters entered
JB> from keyboard? I believe the Tibetan alphabet is not caligraphical
JB> (like chineese), so there should be only few letter (35?). So each
JB> character is typed with single keystroke or do you need some
JB> combinations?
There are 30 letters in Tibetan alphabet. But they are modified in
their graphic representation depending on their position in a
syllable. They are combined from left to right and from top to bottom.
The standard way to type Tibetan is to use Wylie transliteration,
where Tibetan letters are represented by English transliterated form.

When I type, for example, a Tibetan syllable "tshan", first I type "t"
and Tibetan "t" is displayed on the screen. Then I type "s", and it
means I have to delete "t" (by sending a backspace simulated
keystroke) and display Tibetan "ts". Same with "tsh" - "ts" should be
deleted and replaced by "tsh". Then I type "a" and nothing happens,
because "a" has no graphical representation in "tshan" glyph, and at
the end - "n", and Tibetan "n" is added to "tsh", forming the final
syllable.

Best wishes,
 Grisha

[kde-russian] Re: Re[2]: Tibetan input

[Lars Knoll]

Hi Grisha and Juraj,

> JB> we've been consulting these things with Eric Bischoff. Our best
> JB> bet is to try to summarize, what Tibetan characters are NOT in
> JB> Unicode and try to file a proposal to add them there. Unicode
> JB> standard is constantly improving and there should be no problem
> JB> with this.

Are you sure they are missing in the unicode standard? Arial Unicode does not 
have all glyphs that are defined in Unicode. You might want to check with the 
glyph tables on www.unicode.org/charts.

I had a short look at your font, and it doesn't even seem to have a Unicode 
encoding, but uses the latin1 area. I would not advice this, as it will 
interfere with latin text, ie. you will not be able to render latin at the 
same time. After reading through the chapter on tibetian in the Unicode book 
once again, I had the impression the code points presented are good enough 
for rendering tibetian.

It seems to me Qt/KDE could already render tibetian quite ok when using the 
code points in Unicode, as it has support for attaching characters together.

> JB>  Would you like to help this?
> Yes, definitely. Actually, from lgm.ttf font I can see for sure that
> 255 symbols are quite enough for all Tibetan input, including
> punctuation and complex Sanskrit stacks for mantras, that are very
> common in Tibetan texts. However, most of Tibetan letters are combined
> from top to bottom, and sometimes there are four symbols combined to
> give a single Tibetan letter, like HUM.
>
> The existing Tibetan Unicode page was not designed with this feature
> in mind, so it is not applicable for real life, and we should modify
> it or have another codepage assigned.

I think it was. the characters 0x0f90-0x0fbc are for example supposed to be 
stacked below. They have a character property that tells (at least the Qt 
font engine) to stack them below the previous letter. Please read the pages 
describing tibetian in the Unicode standard. If you don't have a recent 
version of the standard (only exists in printed form), tell me and I'll try 
to scan in the relevant pages.

> We can even try to make an 8 bit codepage for Tibetan, because with
> smart design all basic Tibetan characters can be fit into 128 symbols.
>
> JB> Second thing: as I don't have Windows license, I can't try your
> JB> program. Could you please explain how are the characters entered
> JB> from keyboard? I believe the Tibetan alphabet is not caligraphical
> JB> (like chineese), so there should be only few letter (35?). So each
> JB> character is typed with single keystroke or do you need some
> JB> combinations?
> There are 30 letters in Tibetan alphabet. But they are modified in
> their graphic representation depending on their position in a
> syllable. They are combined from left to right and from top to bottom.
> The standard way to type Tibetan is to use Wylie transliteration,
> where Tibetan letters are represented by English transliterated form.
>
> When I type, for example, a Tibetan syllable "tshan", first I type "t"
> and Tibetan "t" is displayed on the screen. Then I type "s", and it
> means I have to delete "t" (by sending a backspace simulated
> keystroke) and display Tibetan "ts". Same with "tsh" - "ts" should be
> deleted and replaced by "tsh". Then I type "a" and nothing happens,
> because "a" has no graphical representation in "tshan" glyph, and at
> the end - "n", and Tibetan "n" is added to "tsh", forming the final
> syllable.

Unfortunately the X11 keyboard handling is 20 years old, and rather hard to 
get additional keyboard with advanced features implemented. This sounds like 
it can unfortunately not be done with a standard X11 keyboard layout, but as 
if you need an input method for it.

I'd propose to first of all concentrate to get text rendering right and then 
see how one can get input to work. I would very much try to use Unicode, as 
it is the standard way all major companies have decided to go for in the 
future, and more or less the only way to keep things interoperable between 
different languages.

Lars