Well, right now it's not interoperable. That's the whole thing. Both 
parties say they do it right yet in the real world it does not work 
although both parties say they do it right. The effect is that it's 
not working because both parties think they do it right.

This has nothing to do with the will to go one step ahead. It's about 
interoperability. If we change it to make it work with Akonadi, it's 
going to break with other clients. Btw, the only response on the RFC 
mailing list interprets the RFC like I do.

Any news on this?

The problem I face right now is that the Horde people say it's 
Akonadi's fault. They don't do it correctly. And the Akonadi people 
say it's Horde's fault. They don't do it correctly.

In the end the result is, that it just doesn't work because neither 
party is willing to go one step ahead and make it work.

Of course, it's coming from Akonadi. ;-)

I still think the example is wrong and contradicts what is written 
earlier in the RFC. Let's see if anything come out of the mailing list 
message. Thanks for writing that message.

This is exactly how the example at the end of section 3.6.2 of RFC 
5545 looks (modulo the PERCENT-COMPLETE). That example gives a 
completed todo with a STATUS:NEEDS-ACTION...

A request for clarification on how a task should be marked as completed:

http://lists.osafoundation.org/pipermail/ietf-calsify/2010-January/002234.html

[14:39] <reinhold> hyper_ch: Regarding your STATUS:COMPLETED vs. 
PERCENT-COMPLETE:100, the rfcs 2445 and  5545 (the successor to RFC 
2445) do not require STATUS:COMPLETED to be set. Actually, it does not 
specify at all how completion should be indicated. There are three 
completion settings: STATUS:COMPLETED, COMPLETED:Date-Time, 
PERCENT-COMPLETE:100
[14:41] <reinhold> hyper_ch: Actually, the example at the end of 
section 3.6.2 shows a completed todo with STATUS:NEEDS-ACTION... See 
http://pastebin.org/75500
[14:41] <reinhold> hyper_ch: That might be an oversight, though.

Re-reading RFC 2445 again, I'm pretty sure that Akonadi is wrong. They 
use the PERCENT-COMPLETE property to indicate the task status instead 
of the STATUS property like Nag does. RFC 2445 says:

Property Name: PERCENT-COMPLETE
Purpose: This property is used by an assignee or delegatee of a to-do 
to convey the percent completion of a to-do to the Organizer.

Property Name: STATUS
Purpose: This property defines the overall status or confirmation for 
the calendar component.
Description: In a group scheduled calendar component, the property is 
used by the "Organizer" to provide a confirmation of the event to the 
"Attendees".

So the difference is that PERCENT-COMPLETE is used by attendees to be 
consumed by organizers, while STATUS is used by organizers to be 
consumed by attendees. Editing and saving a task on the (Nag) server 
is done by the task organizer (or anyone with the same permissions) 
and authoritative for anyone consuming this task. This is *not* an 
update from an attendee sent to the organizer. The RFC even explicitly 
mentions that PERCENT-COMPLETE is per attendee, so that attendees can 
report back different completion states for their parts of the task. 
It is *not* to be stored with the task. How should that work anyway if 
this a property of the VTODO component but can be different for each 
attendee?

After having put that into the rpc.php and into the nag api.php (at 
_nag_import() and _nag_put() ) I finally got this in the log:

BEGIN:VTODO
DTSTAMP:20091112T140441Z
ORGANIZER:MAILTO:email@email.address
CREATED:20091112T075928Z
UID:libkcal-1455987314.377
SEQUENCE:1
LAST-MODIFIED:20091112T140416Z
SUMMARY:test task
STATUS:NEEDS-ACTION
PRIORITY:5
DUE:20091118T230000Z
COMPLETED:20091112T140416Z
PERCENT-COMPLETE:100
END:VTODO

Try adding:

Horde::logMessage($input, __FILE__, __LINE__, PEAR_LOG_ERR);

(or whatever your minimum error log level is) to rpc.php, right before 
the getResponse() call.
Great for them if this has been discussed before, but it doesn't help 
if no one actually reports it back to us.

If I knew how to grab it with a network logger or is there  a way that 
the rpc.php dumps all queries into a log?

As far as I was told in the akonadi channel is, that this has been 
discussed before and that they determined that horde was the culprit.

[Wednesday, 11. November 2009] [13.51:57] <volker|office>       this 
has been discussed here before, IIRC horde ignores the percent-done 
field apparently
[Wednesday, 11. November 2009] [13.52:30] <bbroeksema|offic>     
well...... then go bug horde people =:)
[....]
[Wednesday, 11. November 2009] [14.04:14] <bbroeksema|offic>    well, 
than see volkers comments =:)
[Wednesday, 11. November 2009] [14.04:30] <bbroeksema|offic>    this 
seem to been discussed before
[Wednesday, 11. November 2009] [14.05:47] <hyper_ch>    ok
[Wednesday, 11. November 2009] [14.08:29] <bbroeksema|offic>     
hyper_ch: The best you can do is contacting horde developers and check 
if this is a known issue for your version of the horde server

Only if Akonadi is correctly sending the completion state. Does the 
console show the actually iCalendar object that is being sent to 
Horde, or can you grab it with a network logger?

I currently use Kontact 4.3.3 and I have setup webdav on the Horde server.

In Akonadi I can either chose eiterh "ICal Calendar File" (which uses 
Akonadi directly) or "KDE Calendar (traditional)" (which uses KCal 
IIRC). For either one I set the resource file like:

webdav://www.domain.com/horde/rpc.php/nag/USER/USER.ics

What works:
- I can add new tasks in Kontact and they will get added to horde
- I can delete tasks in Kontact and they will get added to horde
- I can edit tasks and the changes will be transmitted to horde

What does not work:
When check the box in Kontact that the task is done, it transmits it 
to Akonadi (check with Akonadi console) and it vanishes from my 
filtered list (Akonadi sets it as 100%). However this change is not 
being transmitted to horde. In Horde the task still remains as 
"active" / "undone". When then Akonadi syncs again with the Horde data 
it will unset that task and it appears in kontact undone again.

In #akonadi on freenode I was told that it's horde's fault for not 
"accpeting" those changes.