On 18 Jul 2011, at 17:02, Michael Sweet wrote:

> We could certainly introduce Pango or ICU support as an experimental addition 
> in 1.3 and default in 3.0 for X11. The current situation definitely limits 
> how much Unicode can be supported without platform-specific code.

True - though we can do a lot with what we have now... Languages that are r2l 
we can *just about* manage, but as soon as we need to do any bi-dir text or 
complex text layout we run into troubles.

On the other hand, we can cope with most LGC or CJK stuff now I think, and I've 
had a lot of success with using the higher planes music rendering sections.

And we can render things like Cuneiform, Linear-B and Futhork now, which is 
obviously useful for everyday computing needs...

>> - switch to rendering (at least) word-by-word rather than
>> glyph-by-glyph, as this makes it easier to interpret the context of each
>> glyph in the word and composite it correctly.
> 
> You want line (or phrase) rendering; word rendering loses bi-dir text 
> handling.

That would be my preference too - I was worried it might be viewed as too big a 
jump from our current per-glyph approach though.


> Basically, fl_draw would need to be modified slightly to pass each line it 
> formats through Pango (or ICU) to generate the final display sequences.

Yes, that would be good; make fl_draw handle the passed string as an entity and 
composing the string as a whole.

I guess that still allows for the user to process text in smaller chunks by 
passing individual words, or even "characters", to fl_draw if they want to 
bypass the CTL.

>> Is there something specific about number handling for Arabic texts? I
>> don't know but I always assumed, since the number system in Arabic texts
>> is derived from broadly the same Indic sources as the numbers used in
>> modern LGC scripts, that they would be more or less the same - is that
>> not the case?
> 
> It is, unfortunately, not the case...

Ah, right. That's a pity, as I'd hoped it would Just Work...


>> Something got clipped here?
>> I guess "reposition the text on the line" or something?
> 
> Ooops, must have deleted the rest by accident - "This could be used to 
> reposition widgets and change the appearance of widgets."

Oh, I see; so if we determine the text is predominantly r2l for example, we 
could redraw the widgets for a r2l biased layout, that sort of thing? 
E.g. move the buttons in a stock dialog to the lower left rather than lower 
right, and so on? (Do we / should we, do that for r2l text? I don't know...)



> 
>>> 3. Methods to implement localization of strings (perhaps layered
>>> on top of the corresponding OS API, or using FLTK-specific code - I
>>> can contribute a .po/.strings file loader)
> 
> 

> I've done enough wrappers - not a big deal to implement efficiently.

Sounds like we might need that!






_______________________________________________
fltk-dev mailing list
[email protected]
http://lists.easysw.com/mailman/listinfo/fltk-dev

Reply via email to