Patrick send me a huge patch fixing many issues in our current SyncML
implementation.
I looked at the patch briefly and really like it 
As the patch arrived on the day of the 1.4.002 maintenance release and
it’s really huge, I decided to not include it immediately.
I want to import it into trunk in the next days, so it can have some
more review and testing. If it improves our SyncML support as described
by Patrick and there are no problems, we can add it in a few weeks to
the 1.4 branch and have an other maintenance release.
I also asked Patrick about helping to develop and maintain the SyncML
code in eGroupWare, to which he agreed. Therefore I want to add Patrick
as a developer to the project and I’m willing to mentor him, if there’s
no other candidate.
Please object, if you dont agree …
Ralf
Patrick Bihan-Faou schrieb:
Hi Ralf,
Ralf Becker a écrit :
Hi Patrick,
Patrick Bihan-Faou schrieb:
Ralf Becker a écrit :
Hi Patrick,
sure send them to me.
Are you interested in joining the project to help maintain the SyncML stuff?
Yes, I can handle that I think.
good news
I am preparing the patchset for syncml, expect it for later today.
Ok. I’m preparing for the 1.4.002 bugfix and maintainance release, which
is to be released tomorrow.
It would be helpful if you not only provide a patch but also a kind of
list what’s fixed and how you assess the stability of the new code.
Here is my set of patches for syncml. This patchset covers a bit more
though and I left some unrelated stuff because I feel they are important
in areas that approach syncml.
I used svn rev 24462 from the 1.4 branch as the base for the diffs so it
should apply fairly cleanly without any trouble…
In addition to this patch, the files in syncml/sif* should be removed as
they serve no purpose anymore (the code was merged in the other
syncml/[contact|calendar|…]/api.php files).
Here is a quick summary of what this set of patches brings:
- handles the default config for current versions of funambol (i.e. the
scard/stask/snote/scal locations)
- tries to be a bit smarter on how the data content should be encoded
based on what the client specified (sif+base64/vcard, / fragmented or
not, etc.)
- workaround a bug in some versions of funambol, where funambol does not
specify the proper sif type for the type of requested data
- imported patch #117 from egw’s tracker
- make sure that the logs generated by the horde code go to stderr so
they can be view in the webserver’s logs
- everywhere the Content-length HTTP header is set, make sure we do
count actual bytes rather than just “characters” as this may be wrong in
multibyte setups (this covers more than just syncml)
- as much as possible reduce code duplication. For example, the
categories are handled in the parent classes for both the SIF avn VCAL
formats for each type of data (addressbook,infolog,calendar).
- make sure the code can handle more than one categories in each direction
- treat the ‘sony ericsson’ vendor string just like ‘sonyericsson’, the
newer phones apparently have a space in the vendor string… (this
touches some files in the icalsrv as well)
- handle notes: these should now work with everything (funambol or other)
- remove more code duplication: the syncml “api” for the various data
types (calendar, contacts, infolog) is now common for both the vcard and
sif data formats (cf the files that need to be removed)
- handle the “privat” filter in infolog like the “private” filter (some
part of the code use the name without the trailing e)
- imported patch # 267 from egw’s tracker
That’s about it…
Completely unrelated, but potentially usefull:
- add support for “logout” when HTTP auth is used (in the logour.php
file) this is completely unrelated to syncml, but is quite handy… Feel
free to exclude that part of the patchset.
All in all, I have about half a dozen installs of egw with these patches
and with them nobody complains about problems with syncs over SyncML
anymore. There might be some bugs in there, but so far this is quite
stable. These patches have made some users select egroupware (in cases
where syncs with outlook where an absolute must have).
I have noticed one issue, which is probably present in the unpatched
code as well: when a participant is removed from an event, the remote
device is not told to delete the event in question for that participant.
This is because the LUID/GUID map is not used by the SyncML code yet for
calendar items. Note that similar bugs are likely to be present with tasks.
I plan on working further on this code, I’ll try to add support for
event participants in SIF (it is being implemented in funambol 6.5) and
also email sync using syncml. Also I’d like to work on server triggered
syncs, but this is more far fetched…
If you have any question, please contact me.
Don’t pay attention to the svn rev that is indicated in the patch. I
have my own subversion repos to maintain my changes to egroupware: trunk
== my version, egroupware’s version is branches/egroupware.org.
Patrick.
–
Ralf Becker
eGroupWare Training & Support ==> http://www.egroupware-support.de
Outdoor Unlimited Training GmbH [www.outdoor-training.de]
Handelsregister HRB Kaiserslautern 3587
Geschäftsführer Birgit und Ralf Becker
Leibnizstr. 17, 67663 Kaiserslautern, Germany
Telefon +49 (0)631 31657-0
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft® Visual Studio 2005.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
eGroupWare-developers mailing list
eGroupWare-developers@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/egroupware-developers
