I've tacked on a few more patches, spelling out more explicitly how DBus_Property parameters work.

I was just speaking to Marco, and he didn't know that you can tell whether a call is incoming or outgoing by comparing InitiatorHandle to yourself and/or TargetHandle. This should go in there somewhere.

(In reply to comment #2)
> I was just speaking to Marco, and he didn't know that you can tell whether a
> call is incoming or outgoing by comparing InitiatorHandle to yourself and/or
> TargetHandle. This should go in there somewhere.

Isn't Requested a simpler way to do this?

(In reply to comment #3)
> (In reply to comment #2)
> > I was just speaking to Marco, and he didn't know that you can tell whether a
> > call is incoming or outgoing by comparing InitiatorHandle to yourself and/or
> > TargetHandle. This should go in there somewhere.
> 
> Isn't Requested a simpler way to do this?

Yes. I have updated the branch with a new patch explicitly documenting Requested being used for this (and also a patch adding shorthand for org.freedesktop.Telepathy in dbus-ref namespaces).

Marco, would http://people.freedesktop.org/~wjt/telepathy-spec-clarifications/spec/Channel_Type_Streamed_Media.html#description have answered all your questions and filled you with glee? I guess it doesn't mention TargetID (which IIRC you also didn't know about); I think the Channel interface description should have a big table of all the immutable properties every channel has.

(In reply to comment #4)
> I think the Channel
> interface description should have a big table of all the immutable properties
> every channel has.

http://people.freedesktop.org/~wjt/telepathy-spec-clarifications/spec/Channel.html#description

Thanks, this mostly looks good.

> +          from the group. A client may supply a reason when attempting to
> +          remove members from a group with
> +          <tp:member-ref>RemoveMembersWithReason</tp:member-ref>, and are
> +          supplied by the CM when emitting

s/and are supplied by/and reasons are supplied by/ for better grammar?

> Document that locally timed out calls should also be No_Answer.

I think we should also clarify that the CM shouldn't do this unless it has to? Our position on this has generally been that the timeout is policy, therefore it's a UI decision, so if anything gives up, it should be the UI, with RemoveMembersWithReason([SelfHandle], No_Answer).

(I can see that some CMs might not have any choice, if a server or oFono or whatever does the timing-out, though.)

In Gabble, we historically had a 1 minute timeout, which Daf removed in 2009 (late in the 0.7.x series) to fix Bug #21153.

>+            namespace = namespace.replace('ofdT.', 'org.freedesktop.Telepathy.')

I'd prefer this to be anchored at the beginning:

    if namespace.startswith('ofdT.'):
        namespace = 'org.freedesktop.Telepathy.' + namespace[5:]

fixed these, i have, hmm?

Indeed you have. Ship it.

commit 8a41e0f19e28df41ba81e7e887732d0ab5d4495c
Merge: 5492810 df57c2a
Author: Will Thompson <will.thompson@collabora.co.uk>
Date:   Tue Jul 13 15:30:39 2010 +0100

    Merge branch 'clarifications'
    
    Reviewed-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
    Makes-everything-clear-for: Marco Barisione <marco.barisione@collabora.co.uk