opensync-users

[Opensync-users] Problems with syncing SonyEricsson P910i and Evolution

[Stian J. <li...@jo...>]

Hi,

This is only with contacts, have not started trying tasks and calendar
yet :)

I guess you might be wanting some traces if you're gonna look into this.
But instead of attaching loads of traces, I thought I'd just describe
the problems first, and you can ask for whatever you want, if you're
going to look at it :)

All this is with SVN from today, and syncml over http. And I understand
that much (most?) of these bugs/errors might be due to bugs in the
phone. I don't know. But I would like to get to know, and perhaps add a
workaround if possible...

1:
First problem first, when I make a change on the phone, and then sync; I
get: 

Member 2 of type evo2-sync committed all changes.
Member 1 of type syncml-http-server committed all changes.
All clients have written
connection with device has ended
Member 2 of type evo2-sync just disconnected
Member 1 of type syncml-http-server had an error while disconnecting: Timeout while waiting for a reply to message "DISCONNECT"
All clients have disconnected
The sync was successful

But when a change has happened in Evolution, I get the normal:

Member 2 of type evo2-sync committed all changes.
Member 1 of type syncml-http-server committed all changes.
All clients have written
connection with device has ended
Member 2 of type evo2-sync just disconnected
Member 1 of type syncml-http-server just disconnected
All clients have disconnected
The sync was successful

I guess this is due to a stupid phone, but I never had this problem with
Multisyncs syncml plugin (which I had several nights on irc with Bo to
debug :) Besides it makes testing boring, since I have to wait a while
before it times out :P

2:
I create an contact in evo called "name1 name2" (File as "name2, name1".
Puts in a couple of phonenumbers. This contact syncs fine, but on the
phone I get two "fields" below the normal fields were the text "name1
name2" and "name2, name1" are. I think this is because this is some
fields that my phone doesn't really recognize.

(If I choose to "File As: name1 name2" I get "name1 name2" twice
instead)

3:
If I change this contact on the phone, and syncs it, the "Full name" in
Evolution disappears (but "File as" is still correct). Though, this
makes me unable to edit the contact in Evolution without first re-adding
the "Full name". 

4:
When I delete a contact in Evolution, I get:

Member 1 of type syncml-http-server had an error while commiting changes: Unable to add data.
Error writing entry 155 to member 1: Timeout while waiting for a reply to message "COMMIT_CHANGE"
Mapping Write Error: Timeout while waiting for a reply to message "COMMIT_CHANGE"
Error writing entry 156 to member 1: Timeout while waiting for a reply to message "COMMIT_CHANGE"
Mapping Write Error: Timeout while waiting for a reply to message "COMMIT_CHANGE"
Member 2 of type evo2-sync just disconnected
Received an reply to our sync
*** glibc detected *** double free or corruption (fasttop): 0xb3f01388 ***
Aborted

Deleting on the phone works fine, after sync it is deleted in Evo as
well. Just not from Evo...

Well, that was it for now. Hope you have time to look into some of this.
Or at least give me some feedback...

Thanks for you incredibly great work on OpenSync!

Best regards,
Stian

Re: [Opensync-users] Problems with syncing SonyEricsson P910i and Evolution

[Armin B. <arm...@de...>]

Hi Stian,

Stian Jordet wrote:
> Hi,
>=20
> This is only with contacts, have not started trying tasks and calendar
> yet :)
>=20
> I guess you might be wanting some traces if you're gonna look into this=
=2E
> But instead of attaching loads of traces, I thought I'd just describe
> the problems first, and you can ask for whatever you want, if you're
> going to look at it :)
>=20
> All this is with SVN from today, and syncml over http. And I understand=

> that much (most?) of these bugs/errors might be due to bugs in the
> phone. I don't know. But I would like to get to know, and perhaps add a=

> workaround if possible...
>=20
> 1:
> First problem first, when I make a change on the phone, and then sync; =
I
> get:=20
>=20
> Member 2 of type evo2-sync committed all changes.
> Member 1 of type syncml-http-server committed all changes.
> All clients have written
> connection with device has ended
> Member 2 of type evo2-sync just disconnected
> Member 1 of type syncml-http-server had an error while disconnecting: T=
imeout while waiting for a reply to message "DISCONNECT"
> All clients have disconnected
> The sync was successful
>=20
> But when a change has happened in Evolution, I get the normal:
>=20
> Member 2 of type evo2-sync committed all changes.
> Member 1 of type syncml-http-server committed all changes.
> All clients have written
> connection with device has ended
> Member 2 of type evo2-sync just disconnected
> Member 1 of type syncml-http-server just disconnected
> All clients have disconnected
> The sync was successful
>=20
> I guess this is due to a stupid phone, but I never had this problem wit=
h
> Multisyncs syncml plugin (which I had several nights on irc with Bo to
> debug :) Besides it makes testing boring, since I have to wait a while
> before it times out :P

this is a bug on my todo list. i can reproduce it with a phone here so i
 got all the information i need.

>=20
> 2:
> I create an contact in evo called "name1 name2" (File as "name2, name1"=
=2E
> Puts in a couple of phonenumbers. This contact syncs fine, but on the
> phone I get two "fields" below the normal fields were the text "name1
> name2" and "name2, name1" are. I think this is because this is some
> fields that my phone doesn't really recognize.
>=20
> (If I choose to "File As: name1 name2" I get "name1 name2" twice
> instead)
>=20
> 3:
> If I change this contact on the phone, and syncs it, the "Full name" in=

> Evolution disappears (but "File as" is still correct). Though, this
> makes me unable to edit the contact in Evolution without first re-addin=
g
> the "Full name".=20

2 and 3 are probably related. The problem is that evo2 and your phone
store the information in different fields. To solve this i need traces
of both syncs (separate please) so that i can see what exactly the phone
sends and what evo2 sends.

but please, try to sync only a single contact so that the log does not
get too long.

>=20
> 4:
> When I delete a contact in Evolution, I get:
>=20
> Member 1 of type syncml-http-server had an error while commiting change=
s: Unable to add data.
> Error writing entry 155 to member 1: Timeout while waiting for a reply =
to message "COMMIT_CHANGE"
> Mapping Write Error: Timeout while waiting for a reply to message "COMM=
IT_CHANGE"
> Error writing entry 156 to member 1: Timeout while waiting for a reply =
to message "COMMIT_CHANGE"
> Mapping Write Error: Timeout while waiting for a reply to message "COMM=
IT_CHANGE"
> Member 2 of type evo2-sync just disconnected
> Received an reply to our sync
> *** glibc detected *** double free or corruption (fasttop): 0xb3f01388 =
***
> Aborted

ok. for this i need a trace of the sync as well as a gdb backtrace of
the crash (if gdb does not trigger, you can try to use valgrind).

Thanks for your help!

Armin

>=20
> Deleting on the phone works fine, after sync it is deleted in Evo as
> well. Just not from Evo...
>=20
> Well, that was it for now. Hope you have time to look into some of this=
=2E
> Or at least give me some feedback...
>=20
> Thanks for you incredibly great work on OpenSync!
>=20
> Best regards,
> Stian
>=20
>=20
>=20
> -------------------------------------------------------
> Using Tomcat but need to do more? Need to support web services, securit=
y?
> Get stuff done quickly with pre-integrated technology to make your job =
easier
> Download IBM WebSphere Application Server v.1.0.1 based on Apache Geron=
imo
> http://sel.as-us.falkag.net/sel?cmd=3Dlnk&kid=3D120709&bid=3D263057&dat=
=3D121642
> _______________________________________________
> Opensync-users mailing list
> Ope...@li...
> https://lists.sourceforge.net/lists/listinfo/opensync-users
>=20

Re: [Opensync-users] Problems with syncing SonyEricsson P910i and Evolution

[Stian J. <li...@jo...>]

tor, 20,.04.2006 kl. 22.00 +0200, skrev Armin Bauer:
> > 2:
> > I create an contact in evo called "name1 name2" (File as "name2, name1".
> > Puts in a couple of phonenumbers. This contact syncs fine, but on the
> > phone I get two "fields" below the normal fields were the text "name1
> > name2" and "name2, name1" are. I think this is because this is some
> > fields that my phone doesn't really recognize.
> > 
> > (If I choose to "File As: name1 name2" I get "name1 name2" twice
> > instead)
> > 
> > 3:
> > If I change this contact on the phone, and syncs it, the "Full name" in
> > Evolution disappears (but "File as" is still correct). Though, this
> > makes me unable to edit the contact in Evolution without first re-adding
> > the "Full name". 
> 
> 2 and 3 are probably related. The problem is that evo2 and your phone
> store the information in different fields. To solve this i need traces
> of both syncs (separate please) so that i can see what exactly the phone
> sends and what evo2 sends.
> 
> but please, try to sync only a single contact so that the log does not
> get too long.

Attached files 2.txt and 3.txt (which I _think_ is the relevant files.
Even though I have only one contact on both Evo and the phone, the
traces get huge! Guess that's normal. I have done excatly as I wrote
here. 

For 2.txt I added a contact with "Full name: name1 name2" "File as:
name2, name1" and cellphone 12345678 in Evo and synced.

For 3.txt I added an email-address "te...@te..." on the phone, and
synced. Now the contacts "Full name" disappeared from Evo.

I don't know syncml at all, but it seems that Evolution tries to save
two (!) different <FormattedName> (which I'm pretty sure are the ones
the phone doesn't recognize). Then on the sync back, the phone tries to
re-add both <FormattedName>, but obviously with a type that Evolution
doesn't recognize... 

Since I don't know SyncML and vCards, I don't know if those
FormattedName's should be recognized by the phone, but do you think
there could be an option to make Evo not write FormattedName to the
phone, while making OpenSync translate LastName, FirstName to
Formattedname? Or is this spec-breaking? Probably is...

> > 4:
> > When I delete a contact in Evolution, I get:
> > 
> > Member 1 of type syncml-http-server had an error while commiting changes: Unable to add data.
> > Error writing entry 155 to member 1: Timeout while waiting for a reply to message "COMMIT_CHANGE"
> > Mapping Write Error: Timeout while waiting for a reply to message "COMMIT_CHANGE"
> > Error writing entry 156 to member 1: Timeout while waiting for a reply to message "COMMIT_CHANGE"
> > Mapping Write Error: Timeout while waiting for a reply to message "COMMIT_CHANGE"
> > Member 2 of type evo2-sync just disconnected
> > Received an reply to our sync
> > *** glibc detected *** double free or corruption (fasttop): 0xb3f01388 ***
> > Aborted
> 
> ok. for this i need a trace of the sync as well as a gdb backtrace of
> the crash (if gdb does not trigger, you can try to use valgrind).

Here's gdb's output, don't know if you need it. Backtrace is pasted
after that.

(gdb) run --sync test2 --wait
Starting program: /usr/local/bin/msynctool --sync test2 --wait
[Thread debugging using libthread_db enabled]
[New Thread -1212654976 (LWP 823)]
Synchronizing group "test2"
[New Thread -1224787024 (LWP 1047)]
[New Thread -1233179728 (LWP 1048)]
[New Thread -1241572432 (LWP 1049)]
[New Thread -1249965136 (LWP 1050)]
[New Thread -1258357840 (LWP 1051)]
[Thread -1258357840 (LWP 1051) exited]
[New Thread -1266750544 (LWP 1052)]
[Thread -1266750544 (LWP 1052) exited]
received contact dsession
Member 1 of type syncml-http-server just connected
[New Thread -1266750544 (LWP 1061)]
[New Thread -1275143248 (LWP 1062)]
[New Thread -1258357840 (LWP 1069)]
[New Thread -1277166672 (LWP 1070)]
Member 2 of type evo2-sync just connected
All clients connected or error
Received a entry pas-id-4447F2510000008E with data of size 0 from member
2. Changetype DELETED
Member 2 of type evo2-sync just sent all changes
Received an reply to our Alert
Going to receive 0 changes
Member 1 of type syncml-http-server just sent all changes
All clients sent changes or error
[Thread -1277166672 (LWP 1070) exited]
[Thread -1258357840 (LWP 1069) exited]
All conflicts have been reported
Member 1 of type syncml-http-server had an error while commiting
changes: Unable to add data.
Member 2 of type evo2-sync committed all changes.
All clients have written
Error writing entry 160 to member 1: Timeout while waiting for a reply
to message "COMMIT_CHANGE"
Mapping Write Error: Timeout while waiting for a reply to message
"COMMIT_CHANGE"
[New Thread -1258357840 (LWP 1191)]
[Thread -1258357840 (LWP 1191) exited]
Member 2 of type evo2-sync just disconnected
Received an reply to our sync

Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread -1233179728 (LWP 1048)]
0x45494c43 in ?? ()


Backtrace:
(gdb) bt full
#0  0x45494c43 in ?? ()
No symbol table info available.
#1  0xb7ec524a in osync_context_report_success (context=0xb3e4a198) at
opensync_context.c:76
        __func__ = "osync_context_report_success"
        __PRETTY_FUNCTION__ = "osync_context_report_success"
#2  0xb73711ec in _manager_event (manager=0x474e4148,
type=SML_MANAGER_SESSION_FINAL, session=0x8247d30, error=0x0,
userdata=0x808c530) at syncml_plugin.c:294
        __func__ = "_manager_event"
        oserror = <value optimized out>
#3  0xb73355fe in smlManagerDispatch (manager=0x808c670) at
sml_manager.c:450
        __FUNCTION__ = "smlManagerDispatch"
#4  0xb7371328 in _sessions_dispatch (source=0x808d3b8, callback=0,
user_data=0x808c530) at syncml_plugin.c:390
No locals.
#5  0xb7ef4876 in g_main_context_dispatch ()
from /usr/lib/libglib-2.0.so.0
No symbol table info available.
#6  0xb7ef7936 in g_main_context_check () from /usr/lib/libglib-2.0.so.0
No symbol table info available.
#7  0xb7ef7c58 in g_main_loop_run () from /usr/lib/libglib-2.0.so.0
No symbol table info available.
#8  0xb7f10582 in g_static_private_free ()
from /usr/lib/libglib-2.0.so.0
No symbol table info available.
#9  0xb7b8b341 in start_thread ()
from /lib/tls/i686/cmov/libpthread.so.0
No symbol table info available.
#10 0xb7e3a4ee in clone () from /lib/tls/i686/cmov/libc.so.6
No symbol table info available.
(gdb)


Don't know if this helps? Trace is attached as tar.gz for this session,
since I had no idea what you needs.

> Thanks for your help!

Gee, it's me (and all of us) who should thank! I'd be happy to donate
something to this project, of you'd need something!

Best regards,
Stian

Re: [Opensync-users] Problems with syncing SonyEricsson P910i and Evolution

[Stian J. <li...@jo...>]

Hi, Armin.

Armin Bauer wrote:
> ok. i can see the problem now. the cause of the problem is that evo2
> stores the name in 3 ways:
> 
> - the N value, where each name part is separated by a ;
> - The FN (formatted Name) which is something like "firstname lastname"
> - the fileAs attribute which is "lastname, firstname"
> 
> the problem is: which one should i use?
> 
> if i either use FN or fileAs to other one would be lost during the sync
> (or i have to regenerate it in the evo2 plugin)

Good question. It seems the P910 only uses the N value? Is that what 
other phones are doing as well? So perhaps the problem is with Evo2? Or 
is it a phone bug?

Either way, it would be (very) nice with an option to "create" the other 
two fields from the N value on my phone... But you know this best.

> this was a bug in libsyncml. i fixed it now in the dev branch of
> libsyncml. please run:
> 
> svn switch http://svn.opensync.org/libsyncml/branches/6620-branch .
> 
> in the main libsyncml directory to switch to this branch.
> 
> please let me know if this problem is fixed then.


I'll test this after work :)

Thanks!

Best regards,
Stian

Re: [Opensync-users] Problems with syncing SonyEricsson P910i and Evolution

[Armin B. <arm...@de...>]

Stian Jordet wrote:
> Hi, Armin.
>=20
> Armin Bauer wrote:
>> ok. i can see the problem now. the cause of the problem is that evo2
>> stores the name in 3 ways:
>>
>> - the N value, where each name part is separated by a ;
>> - The FN (formatted Name) which is something like "firstname lastname"=

>> - the fileAs attribute which is "lastname, firstname"
>>
>> the problem is: which one should i use?
>>
>> if i either use FN or fileAs to other one would be lost during the syn=
c
>> (or i have to regenerate it in the evo2 plugin)
>=20
> Good question. It seems the P910 only uses the N value? Is that what
> other phones are doing as well? So perhaps the problem is with Evo2? Or=

> is it a phone bug?
>=20
> Either way, it would be (very) nice with an option to "create" the othe=
r
> two fields from the N value on my phone... But you know this best.

yes, i can generate the value from the N field. The problem appears if a
user changed the fileAs attribute in evo2.

lets say someone replaced the fileAs attribute with which was "Doe,
John" before with "John". If we would generate the fileAs attribute
during the sync we would overwrite this change again with "Doe, John".
Dont know if there is a easy solution to this problem. I guess the best
approach would be just to generate these fields and hope not too many
people fill in custom values.

>=20
>> this was a bug in libsyncml. i fixed it now in the dev branch of
>> libsyncml. please run:
>>
>> svn switch http://svn.opensync.org/libsyncml/branches/6620-branch .
>>
>> in the main libsyncml directory to switch to this branch.
>>
>> please let me know if this problem is fixed then.
>=20
>=20
> I'll test this after work :)
>=20
> Thanks!
>=20
> Best regards,
> Stian
>=20

Re: [Opensync-users] Problems with syncing SonyEricsson P910i and Evolution

[Stian J. <li...@jo...>]

fre, 21,.04.2006 kl. 11.59 +0200, skrev Armin Bauer:
> > Either way, it would be (very) nice with an option to "create" the other
> > two fields from the N value on my phone... But you know this best.
> 
> yes, i can generate the value from the N field. The problem appears if a
> user changed the fileAs attribute in evo2.
> 
> lets say someone replaced the fileAs attribute with which was "Doe,
> John" before with "John". If we would generate the fileAs attribute
> during the sync we would overwrite this change again with "Doe, John".
> Dont know if there is a easy solution to this problem. I guess the best
> approach would be just to generate these fields and hope not too many
> people fill in custom values.

But does todays behaviour work fine for everyone except me? If so, could
you please consider have an option forcing only syncing the "N"-value,
and generating the other two from that one? There are at least some P910
users out there...

> > 
> >> this was a bug in libsyncml. i fixed it now in the dev branch of
> >> libsyncml. please run:
> >>
> >> svn switch http://svn.opensync.org/libsyncml/branches/6620-branch .
> >>
> >> in the main libsyncml directory to switch to this branch.
> >>
> >> please let me know if this problem is fixed then.
> > 
> > I'll test this after work :)'

And it works perfect :) Only one big "bug" preventing me from using
OpenSync for real now :)

Thanks.

-Stian

Re: [Opensync-users] Problems with syncing SonyEricsson P910i and Evolution

[Stian J. <li...@jo...>]

fre, 21,.04.2006 kl. 19.02 +0200, skrev Stian Jordet:
> fre, 21,.04.2006 kl. 11.59 +0200, skrev Armin Bauer:
> > > Either way, it would be (very) nice with an option to "create" the other
> > > two fields from the N value on my phone... But you know this best.
> > 
> > yes, i can generate the value from the N field. The problem appears if a
> > user changed the fileAs attribute in evo2.
> > 
> > lets say someone replaced the fileAs attribute with which was "Doe,
> > John" before with "John". If we would generate the fileAs attribute
> > during the sync we would overwrite this change again with "Doe, John".
> > Dont know if there is a easy solution to this problem. I guess the best
> > approach would be just to generate these fields and hope not too many
> > people fill in custom values.
> 
> But does todays behaviour work fine for everyone except me? If so, could
> you please consider have an option forcing only syncing the "N"-value,
> and generating the other two from that one? There are at least some P910
> users out there...

Sorry to nag you again with this, but have you thought of a way to do
this? I'm really eager to start using it for real now :)

-Stian