Re: Where to concentrate effort for Bi Directional support?
On Thu, May 03, 2001 at 12:52:02PM +0300, Shaul Karl wrote:
> A local office of a major international corporation is willing to put effort
> in order to have free software tools to support Bi Directional languages under
> This is probably mostly concerned Arabic and Hebrew. (Are there other
> languages who will directly benefit from such support?)
do not forget farsi, ofcourse.
> They have made it clear that they are actually interested to work on libs or
> other tools of more fundamental nature, as opposed to end user applications.
> What they currently looking for is a list of things from which they can choose
> one or two that they prefer to work on. Can anyone compile such a list? A
> single item maybe? Alternatively, would you say that all the basic support is
> there and if there is a problem for those languages speakers then it is mostly
> with the user level applications?
the problem is more fundamental than user-level applications. the first
two candidates coming to mind are the QT toolkit and the GTK toolkit.
QT 3.0 is rumored to have bidi support, and GTK already has pango, but
i do not know what current level of hebrew it supports, as most gnome
applications have not as yet integrated the new GTK with pango support.
both projects would probably be more than happy to recieve help on this.
the console support however, is truly lacking. i do not see an easy way
of supporting bidi in the console, aside from perhaps screen reversal,
which i have seen in some 3270 emulators. i do not know if it is part
of the standard, or just an implementation quirk in the emulator. anyhow,
a port of any console toolkit (slang/newt/curses) to support hebrew would
be very appreciated. ofcourse, user level applications will have to follow
for the initiative to succeed.
> What would you say about having a field in terminfo that will default to
> left-to-right but might be changed to right-to-left direction?
it'd probably be a bit hackish to implement it this way, if i got the idea.
the right way (imho) would be to revive the linux console project
(see their sourceforge page) to handle unicode and bidi.
> How complex would it be to have basic tools like grep deal with unicode
> without having to quote the exact byte each letter is composed of?
> I have not followed closely the bf thread here but can you make any
> suggestions that are based on this thread?
if all you do is reverse the display terminal, you do not need to concern
yourself with bidi issues. it's a crude hack, but terminals and textmode
in general do not seem to like bidi (anyone remembers hebrew on dos?)
the linux kernel supports unicode to some level, and (afaik) gnu utils take
advantage of that (textutils, fileutils etc). more work to perfect this
support would be more than welcome, however.
i have barely seen any discussion on bidi issues in the bf thread,
> Thank you.
> Shaul Karl <firstname.lastname@example.org>
i bet you've seen this sig somewhere before...
===================== Gil Bahat =====================
================ email@example.com ============
====== world leader in wasting bytes since 1981 =====