[sync] Re: SyncML update

[Johny]

Anthony & Chuck,

Got the hooks.php.dist update, and copied it across to the hooks.php 
dir. Also, did a complete cvs update on both framework and horde, now I 
get this issue when testing the link from Firebird;

Fatal error: Failed opening required 'SOAP/Server.php' 
(include_path='.:/usr/share/pear') in /usr/share/pear/Horde/RPC/soap.php 
on line 43

It goes and invokes the Soap handling by default, not SyncML as the P800 
does.

There are no files in /tmp/sync yet, it doesn't get that far although I 
had those last time I tested this.

The 302 is weird, now it reports 200 as I'd expect, but doesn't seem to 
get anywhere from the call to rpc.php, so I'm adding a touch of 
debugging there to see what is happening now, hoping you guys will sort 
out the SOAP handling :)

For info on 302;
http://www.w3.org/Protocols/rfc2616/rfc2616-sec10.html



:)J

Anthony Mills wrote:
> Are you looking at the auth in SyncML, or in RPC.  SyncML uses an 
> authorizer in State.php line 269.  Look for some files in '/tmp/sync'. 
> The *.xml files are the syncml packets.  syncml_client_*.xml is sent 
> from your phone, syncml_server_*.xml is sent from the server.  The 
> server should be sending back data codes of 401 or 212.  The former code 
> means you logged in.  I'm not sure why your getting the 302 in apache, 
> it should be 200.  Do you known what 302 is in http?
> 
> Anthony
> 
> Johny wrote:
> 
>> Set the auth mode to 'auto', still no luck...
>>
>> suggestions very welcome!
>>
>> :)J
>>
>> Johny wrote:
>>
>>> Fixed, I'd completely forgotten about that test page, extremely 
>>> useful, ta! :)
>>>
>>> Now the auth is stumping me :( I have two auth modes on my phone, 
>>> namely an HTTP auth and also a server auth. The first is (as far as 
>>> my testing goes) for HTTP only and not invoked here, the second is 
>>> the key one.
>>>
>>> However, the first one doesn't seem to get invoked either, the apache 
>>> access_log;
>>>
>>> 192.168.1.141 - - [31/Jan/2004:20:19:19 +1000] "POST 
>>> /horde-cvs/horde/rpc.php HTTP/1.1" 302 0 "-" "EPOC SyncML Client"
>>>
>>> The Horde log for the same event;
>>> Jan 31 20:19:20 HORDE [debug] [] User  does not have READ permission 
>>> for horde [on line 637 of 
>>> "/var/www/html/horde-cvs/horde/lib/Registry.php"]
>>>
>>> I added some logging to the 'login' driver to check that it is 
>>> invoked, and I can see it being invoked when using Firebird and the 
>>> login page there, but not when using my phone :(
>>>
>>> Suggestions? Is there an easy way to disable auth so I can test further?
>>>
>>> :)J
>>>
>>>
>>>
>>>
>>> Anthony Mills wrote:
>>>
>>>> Try:
>>>> pear install Log
>>>>
>>>> from a command prompt.
>>>>
>>>> You will need to run 'pear install *' for many packages.  Also, use 
>>>> http://192.168.1.254/horde-cvs/horde/test.php.  It will you setup 
>>>> horde.
>>>>
>>>> Anthony
>>>>
>>>> Johny wrote:
>>>>
>>>>> Ta mate, I've just upgraded my dev server to RH9, got the latest 
>>>>> version of PEAR onto it and at last I got the framework 
>>>>> 'install-packages.php' to work :)'
>>>>>
>>>>> Running with the very latest code from CVS for both framework and 
>>>>> horde I get the following error which I think is a tad too 
>>>>> fundamental for me to figure out, basically it seems Log.php has 
>>>>> gone amiss?
>>>>>
>>>>> I had a look using chora in the repository and couldn't find it, 
>>>>> also had a look in the PEAR and horde dirs but couldn't find a copy...
>>>>>
>>>>> Any advice?
>>>>>
>>>>> :)Johny
>>>>>
>>>>> URL (with Firebird for testing purposes);
>>>>> http://192.168.1.254/horde-cvs/horde/rpc.php
>>>>>
>>>>> Returns;
>>>>> Warning: Failed opening 'Log.php' for inclusion 
>>>>> (include_path='.:/usr/share/pear') in 
>>>>> /var/www/html/horde-cvs/horde/lib/Horde.php on line 3
>>>>>
>>>>> Notice: Use of undefined constant PEAR_LOG_DEBUG - assumed 
>>>>> 'PEAR_LOG_DEBUG' in /var/www/html/horde-cvs/horde/lib/Registry.php 
>>>>> on line 637
>>>>>
>>>>> Fatal error: Undefined class name 'log' in 
>>>>> /var/www/html/horde-cvs/horde/lib/Horde.php on line 74
>>>>>
>>>>>
>>>>> Anthony Mills wrote:
>>>>>
>>>>>> I checked in all my code.  It should do two things, if you put an 
>>>>>> invalid password (the authentication uses horde now) it should 
>>>>>> send a packet back telling you so.  If you send a valid password, 
>>>>>> it should send some content back (this may not work).  Right now 
>>>>>> there is a problem with Encoder from XML_WBXML, I don't know if 
>>>>>> your phone has to have WBXML, if it does you can check out 
>>>>>> libwbxml and use it's wbxml2xml and xml2wbxml utilities to test 
>>>>>> the packets, they get stored in '/tmp/syncml' by default.  I 
>>>>>> suspect the WBXML code is messing up on the switch between URIs.  
>>>>>> If you get stuck, email me, I'll be happy and able (that is if I'm 
>>>>>> not sleeping) to point you in the right direction.  I have too 
>>>>>> much work this week to mess with code, but email is doable.  I can 
>>>>>> even irc for short periods if that would help.
>>>>>>
>>>>>> Thanks for your help,
>>>>>>
>>>>>> Anthony
>>>>>>
>>>>>> PS I updated encode-decode.phpt in XML_WBXML/tests, unfortunately 
>>>>>> it works, but wbxml2xml does not.
>>>>>>
>>>>>> Johny wrote:
>>>>>>
>>>>>>> Anthony,
>>>>>>>
>>>>>>> I'll have time to have a look again this weekend, testing from my 
>>>>>>> phone, so if you check stuff in I'll be happy to test :)
>>>>>>>
>>>>>>> I did have some funny errors last time I tried, didn't get time 
>>>>>>> to check them out properly, but will do so this time round and 
>>>>>>> report.
>>>>>>>
>>>>>>> I'm ahead of you (most likely, I'm in Oz), so if you have a 
>>>>>>> chance at checking in something which works I'm happy to check it 
>>>>>>> out from CVS and test. Just email the list and I'll get it :)
>>>>>>>
>>>>>>> Cheers!
>>>>>>>
>>>>>>> :)Johny
>>>>>>>
>>>>>>> Anthony Mills wrote:
>>>>>>>
>>>>>>>> I just wanted to let everyone know I did not fall off the face 
>>>>>>>> of the earth, I just found a bug.  It seems that my WBXML 
>>>>>>>> encoder does not work.  I am using the wbxml2xml (from libwbxml) 
>>>>>>>> tool to test with.  It fails with a cryptic message.  I got the 
>>>>>>>> server to accept the username and password, which is good, but I 
>>>>>>>> can't send a valid response back. When I finish this we should 
>>>>>>>> be back on track.  I'm going to try to check in everything I 
>>>>>>>> have  so some one can take a look at if they want.
>>>>>>>>
>>>>>>>> Thanks,
>>>>>>>>
>>>>>>>> Anthony
>>>>>>>>
>>>>>>>
>>>>>>
>>>>>
>>>
>