[Bug 7150] Implement Arabic shaping

[wine-bugs at winehq.org]

http://bugs.winehq.org/show_bug.cgi?id=7150


alsadi <alsadi at ojuba.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |alsadi at ojuba.org




--- Comment #13 from alsadi <alsadi at ojuba.org>  2009-10-26 11:57:46 ---
> Muayyad, wine developers don't grab patches from bugzilla, see
my beloved brother, as you know I send patches to several project (anaconda,
publican, gnome-xsl, packagekit, pulseaudio, fedora, ...) and they all accept
my patches through bugzilla

could you please please continue with them if I were too busy.

> * Please keep the spacing of your code consistent with the rest of the file (4
space indentation, spaces around binary operators, etc).

my patches only add lines so put the code I added into a file then call a
indent tool over it with the right arguments winehq wants, that's not a big
deal  :-) 
BTW: what are those argument (so that I would do this step if I send patches to
winehq in the future)

> * You don't need to include your email address in the code itself (as the Wine
Git repository tracks authorship information).

feel free to remove it (I grant you the permission, in case if your lawyer is
worried)

* Can you please use slightly more descriptive variable names than p, c, cn,
ch, jx, jcn, jpc, j, i and ix?

it's a complicated optimized algorithm, in which variables could be reused,
but since those variables are locally used within my function only they don't
worth commenting or being long (as in kernel coding style)

but don't worry here are the meanings
p-> point that moves through the text being processed
c-> current char
jcn-> should we join current with next?
jpc-> should we join previous with current?
j-> point to the some element in joins array
jx -> an element in ArabicShapingData joining data

* Is it possible to write conformance tests for this? (If so, please include
some, as the patch will be much more likely to get in.)

I'm a native speaker and it works, we ship the patched version in our distro (
ojuba linux) and we had offered several patched version in parallel to all wine
updates in fedora 11

we have not received bug reports related to this patch
(multi line editing is due to bidi and it's independent of this patch, small
fonts a config issue, and wine renders chars that supposed to be invisible like
U+2xyz)

regarding non-Arabic text, it can be proved that it will not be affected by
this patch because of if no chars in the range of ArabicShapingData, skip will
be equal to True and the second shaping loop which will do the actual shaping
will be skipped.

> * You may find it easier to manage and generate your patches with Git.
this patch only add lines that are independent of the code, I'm certain that
can be applied cleanly as-is without any extra effort.

please try it, and inform me if it did not.

-- 
Configure bugmail: http://bugs.winehq.org/userprefs.cgi?tab=email
Do not reply to this email, post in Bugzilla using the
above URL to reply.
------- You are receiving this mail because: -------
You are watching all bug changes.