Re: [Opensync-users] SyncML over bluetooth/obex again

[Martin S. <mar...@hi...>]

More comprehensive patches including server alerts compliant to SyncML
1.1 are now in ticket #146. The problem with the idle command I don't
understand either. I will look at it again this weekend.


Am Donnerstag, den 02.02.2006, 02:12 +0100 schrieb Armin Bauer:
> Hi,
> 
> thanks for the patches! i will integrate them in libsyncml branch and
> the plugin once i find some time (probably means: after my exams :)
> 
> Martin Schulze wrote:
> > I've had a look at the libsyncml-threaded version this evening. From the
> > sparse information that I could get, this seems to be the actively
> > maintained branch. 
> 
> That is correct. All development is currently done in the threaded-branch.
> 
> > Attached are patches to this branch and the
> > sycml-plugin from trunk.
> > 
> > With these patches applied to current svn, sycml-plugin compiles and
> > runs with the libyscml-threaded version. There is no ChangeLog, so I
> > have enclosed a short description of the changes below.
> > 
> > msynctool still doesn't work with my mobile phone. It doesn't seem to
> > react on the SAN. I gather so much that SAN requires syncml 1.2 (at
> > least the way it is currently implemented in libsyncml) and the SGH-D600
> > only support syncml 1.1.
> 
> SAN was not defined prior to syncml 1.2 so there is no chance that a 1.1
> phone can make any sense of it. in syncml 1.1 you just have to send a
> normal alert to the phone to trigger the synchronization. I havent
> tested yet if this is correctly done (at least not with a real phone).
> 
> > 
> > Regards,
> > 
> >   Martin
> > 
> > _____________________________________________
> > 
> > libsyncml-2006_01_29.diff:
> > 
> > 	libsyncml/sml_queue.c: Rename the callback functions. Before, the
> > callback functions from libopensync/osengine/osengine_queue.c were
> > called on my testing platform (leading to SEGFAULT).
> 
> right. i think we should declare them static so that naming conflicts
> like this cannot happen (or use the proper prefix)
> 
> > 	libsyncml/sml_transport.c: Fix typo in the documentation.
> > 	libsyncml/sml_manager.c: Enhance _smlManagerDataHandler() to handle to
> > only event type SML_TRANSPORT_EVENT_DATA. TODO: Probably, events of type
> > SML_TRANSPORT_EVENT_CONNECT_DONE should also be forwarded to
> > manager->eventCallback(). However, there is no SML_MANAGER* macro for
> > this, yet.
> > 	smlManagerSessionRemove() now calls manager->eventCallback() with
> > SML_MANAGER_SESSION_END.
> > 	smlManagerNew(): Fix typo.
> > 	libsyncml/sml_support.c: The idle callback doesn't get called when a
> > new thread is started. Work around: _start_main_loop().
> 
> i dont really understand why this is necessary. i used this code quite
> often now and it worked very well.
> 
> > 
> > syncml-plugin-2006_01_29.diff:
> > 
> > 	src/syncml_plugin.[hc]: Make it work with the libsyncml-threaded branch
> > of libsyncml. Use smlManager to manage sessions. I'm not sure whether
> > SmlDsSession is used correctly in batch_commit().
> 
> looks ok from the very first glance. the idea is that the syncml session
> is used to control the commands that are sent and received. it also
> handles stuff like the fragmentation. then there might be one or more
> datasync session inside the syncml session where each ds session is
> responsible for synchronizing a single format. so a synchronization
> looks like this:
> 
> -> the syncml session is received
> -> the ds servers get called and return the ds session(s)
> -> the work is done on the ds sessions (like requesting data, sending
> changes etc).
> -> once each step is called for all ds sessions you call smlSessionFlush
> to flush all remaining commands in the syncml session
> 
> Best Regards,
> Armin Bauer
> 
> > 	src/syncml-obex-client: Rename 'path' to 'url'. Add new parameter
> > 'port'.
> >