Unicode support for GG protocol

[GoogleCodeExporter]

Hi,

some time ago I modified the GG protocol plugin to support UTF-8 extensions 
added by new protocol version and fixed the plugin to work correctly with 
Unicode enabled.

I've been using the modified version since that time without any major 
problems, so I decided that it's the right time to share my code. The patch is 
pretty big and some changes might be rather dirty, but as I said, it is 
confirmed by several people that it works without any major problems.

Anyway, I'm pretty busy with other things and I'm not going to maintain/update 
this patch, so the best solution would be to someone adopt it or even include 
into the main source tree.

I'm attaching a patch based on SVN revision 13151.

Best regards
Tomasz Figa

Original issue reported on code.google.com by tomasz.f...@gmail.com on 7 Dec 2010 at 2:23

Attachments:

• gg_unicode_20101207_rev13151.patch

[GoogleCodeExporter]

Original comment by sami%mir...@gtempaccount.com on 7 Dec 2010 at 2:32

• Changed state: Assigned
• Added labels: Component-Protocol-GaduGadu, Type-Patch
• Removed labels: Type-Enhancement

[GoogleCodeExporter]

Well, your patch is incorrect. GG plugin uses 2-in-1 method, i.e. the same 
exact dll used for ANSI and Unicode miranda, this what gg_a2t and similar 
functions do, and you removed them... If you want your patch to be considered, 
please revise the patch, remove all mir_a2t (as this willl never workk right in 
gg) and put gg_a2t back.

I am closing this until revised correct patch provided.

Original comment by borkra on 7 Dec 2010 at 3:22

• Changed state: Invalid

[GoogleCodeExporter]

Excuse me, but this work perfectly fine for me and at least 3 other people and 
there is no reason it shouldn't work. 

There is no way you can provide a single DLL for both Miranda version without 
completely removing Unicode support (or replacing it with translation from 
Unicode to ANSI, which is just the same, because it strips any extra characters 
unavailable in current charset, ie. Win1250).

Also I don't see a reason why mir_a2t wouldn't work. It translates from ANSI to 
Unicode when Unicode is enabled or just duplicates the string in the other 
case. This is not my business but IMHO this is not my solution which is wrong, 
but yours.

Original comment by tomasz.f...@gmail.com on 7 Dec 2010 at 3:28

[GoogleCodeExporter]

@borkra

You wrong, ANSI version not provide receiving and sending Unicode data. GG 
Protocol from GG 8 version support this and tomasz.figa patch work for me 
without problems. With this patch I can set Unicode status and send/receive 
Unicode data, without it a got scrambled text and/or missing characters when 
using Unicode. This should be re-opened and patch should be merged to trunk.

Original comment by tapcio on 7 Dec 2010 at 3:51

[GoogleCodeExporter]

1. You obviously did not try to use your stuff with ANSI Miranda.
2. mir_a2t will not work because GG compiled once without UNICODE flag for 
release (VC2008 VC2010 projects that nobody uses, so I do not know what they 
set to). So as a result mir_a2t will be always mir_a2a 
3. 
> (or replacing it with translation from Unicode to ANSI) 
And this is exactly what gg_xxx functions do, they look on Miranda version info 
and perform Unicode conversion if needed.

> which is just the same
No, it's not, because we do it other way.

4. tapico, I am not talking ANSI version of Miranda which you obviously do  not 
use. And yes ANSI version of Miranda should support Unicode, because current 
locale different for different people. And version of GG protocol is 
irrelevant. I am talking here about Miranda version.

Original comment by borkra on 7 Dec 2010 at 5:31

[GoogleCodeExporter]

1. I think that I also tested my changes with ANSI version of Miranda, but this 
was pretty long time ago. So I may recheck that. Anyway, I see no reason why it 
wouldn't work. 
2. Is there anything wrong with mir_a2a (=== mir_strdup)? It just doesn't 
perform any conversion.
3. Even forgetting that this is less effective (checking a condition every 
gg_x2y), it is wrong because there are many Miranda APIs taking xxx_TCHAR 
(resolving to 0 on !UNICODE and xxx_WCHAR on UNICODE, since xxx_WCHAR variant 
is unsupported when UNICODE is disabled) flags and changing the behavior of 
passed parameters interpretation, without which you couldn't pass an UTF-8 
encoded message without any loss caused by charset conversion.
4. With Unicode version of Miranda and its plugins, you could forget about any 
locale differences.

Original comment by tomasz.f...@gmail.com on 7 Dec 2010 at 6:48

[GoogleCodeExporter]

Hi guys,

I had already started working on GG transition to Unicode some time ago. The 
work is almost finished now, I just postponed the patch because I want to fix 
file transfers first (which also needs to be Unicode). So don't worry, GG will 
be Unicode and will support UTF-8, just give me some time to fix some things.

Original comment by dezred on 7 Dec 2010 at 7:14

[GoogleCodeExporter]

Original comment by sami%mir...@gtempaccount.com on 7 Dec 2010 at 8:08

• Changed state: Started

[GoogleCodeExporter]

@tomasz.figa
1. The problem is not that it can not work the way you did it, but that is not 
the style that was chosen by GG plugin, and when you are providing the patch 
you need to match the style chosen.

2. Yes, with Unicode you can forget about local differences, but we still 
provide ANSI Miranda version and there we cannot forget about it...

Original comment by borkra on 7 Dec 2010 at 9:44

[GoogleCodeExporter]

Still providing ANSI Miranda is for me silly thing, only add developing effort 
and nothing else. Almost no one is using Windows 9x systems. Even Windows XP is 
not dominating, right now most popular is Windows 7.

Original comment by tapcio on 20 Dec 2010 at 11:05