Hi Tavis, I think that Pango has its own shapers for Bengali and other Devanagari based scripts. Right now Abi has its own shaping engine that does not use Pango - we currently use XFT2 on Linux to draw the necessary text.
I'm no i18n expert by any stretch of the imagination, but I *think* this is what's going in here. Cheers, Dom --- Tavis Barr <[EMAIL PROTECTED]> wrote: > > Hi everyone. > > I just installed AbiWord 1.1.3 from the RPMs and I > have to say I'm very > impressed by the progress. Congratulations. > > I'm trying to get the Bengali input method to work > under gtk+. I use a > plugin called imbeng, available from: > > http://sourceforge.net/project/showfiles.php?group_id=43331 > > I fire up AbiWord under and Indian Bengali locale > (LC_ALL=bn_IN.UTF-8 > abiword-2.0 &) and change the font to a Bengali font > (Mukti in this case > but I don't think it matters). > > The Bengali input method is correctly chosen by > default (i.e., if I > right-click in the text entry area, it's selected > under "Input Methods") > and when I start typing, the appropriate Bengali > characters come out. > This already means that it's at least half-working, > because the imbeng > program is responsible for turning multiple > keystrokes into > multiple-keystroke characters. > > However, the characters don't glyph properly when I > display them. There > are two points here and I'll try to give some > background. > > Bengali (and other Devanagari-based scripts) work in > a system where each > syllable in a word is represented by a glyph for a > consonant (or a > single glyph for a string of consonants if they are > pronpounced > together, such as "str" or "pl") and an attached > glyph for the vowel > that follows that consonant. (Words that start with > a vowel get a > separate starting-vowel glyph.) Depending on the > vowel, sometimes the > vowel sign comes to the left of the consonant, > sometimes to the right. > For example, "e" and "i" come to the left of the > consonant in Bengali, > "a" to the right, "u" below, and "o" on both sides. > > So, for example, if I type the Bengali word "sneho" > (meaning > "affection"), I should expect to get a glyph for > "sn", and then the > glyph for "e" which in this case comes to the left > of the glyph for > "sn." > > Instead, I get the glyph for s followed by the glyph > for n, and the "i" > to the right of it all. > > One thing that makes me think that this is an issue > of rendering and not > input: Each of the consonants (s and n), when it is > displayed, is > followed by a special Bengali punctuation marker > underneath that sort of > means "don't treat the next character as a separate > syllable." So the > proper way to render appears to involve combining > consonants into one > glyph when they are followed by this marker. > > Currently, gedit 2.0 renders this input method > correctly. > > I also suspect that if this problem exists for > Bengali, it probably > exists for other Indian languages. > > I'm sorry I don't know more about the internals to > be able to suggest > precisely what's wrong. I hope this note is helpful > anyway in pointing > out a bug. For more information on rendering > Bengali, you might wish to > visit the Bengali Linux project at > www.bengalinux.org and talk to one of > the developers there, who may at least know how it > works on gedit (which > is their reference program). > > Although I'm not familiar with the programs, I can > and do code; let me > know if there is any way I can assist with this. > > > Thanks, > Tavis > __________________________________________________ Do you Yahoo!? Yahoo! Shopping - Send Flowers for Valentine's Day http://shopping.yahoo.com
