Branch is nearly ready. The only issue I'm facing is that an enum has been extended changing significantly the resulting error codes.

As Call was unstable, I'm temped to simply set the error as they are in the new Spec, and leave it to the UI to support the new enum values. Does that make sense ?

Sounds good to me. Calls logging was experimental and had to be manually enabled so I don't really care if we break old logged calls.

Branch ready for review.

The diff's kind of hard to read so I only half looked at it for call-channel.c and just looked at the new file as is.

>  param_spec = g_param_spec_int ("end-reason",
>      "End Reason",
>      "Reason for wich this call was ended",
> -    0, G_MAXINT, TPL_CALL_END_REASON_UNKNOWN,
> +    0, G_MAXINT, TP_CALL_STATE_CHANGE_REASON_UNKNOWN,
>      G_PARAM_READWRITE | G_PARAM_CONSTRUCT_ONLY | G_PARAM_STATIC_STRINGS);

You could use TP_NUM_CALL_STATE_CHANGE_REASONS instead of G_MAXINT here if you wanted?

> - reasons[priv->end_reason],
> + _tpl_call_event_end_reason_to_str(priv->end_reason),

Missing space before the bracket; that happens a couple of times I think.

>  if (contact == NULL)
>      contact = tp_connection_get_self_contact (con);

Should be two spaces instead of one.

Do the calls to dbus_g_object_register_marshaller need to stay?

Other than those minor points it looks basically fine. I don't know much about the logger but I'm sure it'll be fine.

Fixed and merged.