Re: Re: Arabic vim patch (was - BAD BUG)

["Antoine J. Mechelynck" <antoine dot mechelynck at belgacom dot net>]

Bram Moolenaar <Bram at moolenaar dot net> wrote:
> Antoine J. Mechelynck wrote:
[...]
> > Looks like I'm going to play the guinea pig then? Well, this one
> > has the glyphs. Thanks Bram for pinning down the guilty patch. Now
> > if that patch's (303) functionality is still needed, the problem
> > remains of how to have it not break the displaying of Arabic glyphs
> > (and maybe others, which I didn't test) in UTF-8.
>
> Well, since we don't seem to make progress in actually solving the
> problem, I'm trying a brute force method: Don't use ETO_PDY when
> 'rightleft' is set.  This won't work if you have two windows, the
> current one has 'norightleft' and the other with 'rightleft' is
> redrawn (CTRL-L).  But it's better than nothing....
[...]

Note: I don't know if it's related with ETO_PDY or not, but the gvim 6.1.320
(including patch 303) refused to display Arabic glyphs whether or not
'rightleft' was set. Actually I launched it in 'norightleft' mode on a
sample Arabic text file found on the Cream project, on the arabic.txt
helpfile which has a few Arabic characters near the end, and on HTML files
of mine with Arabic text (and of course Latin tags), always with the same
result (no Arabic glyphs). Identical vimrc, gvimrc, etc -- when the patch
303 was removed, with no other changes, the glyphs suddenly reappeared in
both 'rightleft' and 'norightleft' display modes. My file was UTF-8 with
BOM, 'enc' 'tenc' 'fencs' and 'fenc' were set (the latter one automagically)
to values corresponding to that situation, and 'guifont' was
Courier_New:h20:cDEFAULT -- maybe I should have tried cARABIC but didn't;
cDEFAULT didn't bother the -p303 versions.

OTOH the 6.1. "earlier than 300" (I'm not sure of the exact patchlevel)
without the Arabic patch did display Arabic glyphs, even if only in their
"isolated" shapes. The versions including patch 303 displayed Latin accented
letters, or hollow rectangles. A layman's opinion would be that they seemed
to "forget" how to fetch the proper glyphs for Unicode codepoints
represented by several bytes in the UTF-8 internal representation. After I
send this mail I'm someday gonna try a +p303 version on a Latin/Cyrillic
UTF-8 file (part of a "Russian-French dictionary" project that I have). If
by extraordinary it does display Russian characters I'll keep you posted. Or
if any of you wants a copy of that file you're welcome to it: ask, and it
shall be given unto you.

As for CTRL-L, doesn't that apply first and foremost to the current window
(the one with the active cursor)? If it didn't, and now does, we can always
do :windo redraw can't we? Or maybe I mised something...

Best regards,
Tony.