Re: [Evolution-hackers] e_cal_new_system_calendar() -> creates a new calendar each time?

[Patrick Ohly <patrick ohly gmx de>]

On Do, 2011-04-07 at 10:47 +0200, Patrick Ohly wrote:
> Hello!
> 
> I noticed that in 2.32/MeeGo, e_cal_new_system_calendar() always creates
> a new calendar, although there is already one.
> 
> It is defined in gconf as:
> 
> <group uid="1300454894.7178.0@pohly-mobl1" name="On This Computer"
> base_uri="local:" readonly="no"><source
> uid="1300454894.7178.4@pohly-mobl1" name="Personal"
> relative_uri="system" color_spec="#BECEDD"/></group>
> 
> The sequence of events is this:
>      1. e_cal_new_system_calendar()
>      2. e_cal_new_from_uri("local:system", ...
>      3. get source list
>      4. search_known_sources() by comparing e_source_peek_absolute_uri()
>         against "local:system"
>      5. no source found, create anew
> 
> Step 4 fails because there is no absolute URI:
> 
> (gdb) p source->priv->absolute_uri
> $11 = (gchar *) 0x0
> (gdb) p source->priv->relative_uri
> $12 = (gchar *) 0x8079450 "system"
> 
> Therefore e_source_peek_absolute_uri() returns NULL and the comparison
> fails.
> 
> What is the root cause for this issue, and how should it be fixed?

absolute_uri is taken from the "uri" property, so adding that to the
gconf sources works around the issue.

Perhaps dump_common_to_xml_node() and the code after
/* do not store absolute uris for local:system sources */
is to blame?

Or if "uri" is correctly not set, then
e_cal_new_system_calendar()/e_cal_new_from_uri() must be fixed to cope
with that?

One more oddity: on the client side, e_source_get_uri() for the system
calendar returns "local:/system". What is the correct absolute URI for
the local system calendar?


-- 
Bye, Patrick Ohly
--  
Patrick Ohly gmx de
http://www.estamos.de/