Re: [Opensync-users] Samsung SGH-D600: Request not successfull: 83

[Markus W. <ma...@mw...>]

On Wednesday 21 June 2006 00:01, Martin Schulze wrote:
> I tried to get the very same phone working with Opensync some time ago.
> After introducing a lot of hacks, I got some communication working but I
> couldn't proceed due to lack of time.
>
> There are several issues with the SGH-D600:
>
> 1) Like Stefan reported, an AT-Command has to be sent before a
> syncml-session can be started ("AT+CPROT=0\r" did the job for me).

Hm, i guess that could be realized, though communication on another bt channel 
(serial chat?!) has to be done in advance...

> [...]
> 3) Only syncml version 1.1 is supported. That means that the
> syncronization cannot be triggered by sending a SAN-Notification.
> Instead, a server alert (type SML_ALERT_TWO_WAY_BY_SERVER) needs to be
> sent in client_connect() (syncml-plugin.c).
> I don't know the current status of the code base, but back in February I
> needed to change the whole logic of how a sycml session is started in
> opensync.

Rumors are, that 1.1 is now in the code base and ready for testing. 

> 4) Furthermore, there are some weird restrictions in the xml protocol.
> Some headers are needed or forbidden - I don't remember the details and
> gave up somewhere at that point.

> I'm sorry that I never managed to get a patch ready. The problem is that
> basically all the changes are work-arounds for this obscure
> OBEX/syncml-client. As such they must only get active if this client is
> identified. In order to avoid contamination of the codebase with a bunch
> of hacks, opensync needs some mechanisms to guess the sycml client and
> to activate such workarounds on demand.

Unfortunately I am quite limited on time, too. Furthermore I am not very 
talented when it comes to Ccoding or/and understanding what already has been 
done and how obex,bt,syncml and all play together. (Easy readings to get 
started anywhere?!). Maybe if this knowledge is there, I cannot hold back and 
start hacking :-p
But in anycase; I'd be willing to test, give feedbacks and debugs...

BTW: I find the idea of a pluggable "bug-work-around" mechanism very charming.

m