Re: [Evolution-hackers] Isolated unit test status report

[Tristan <tristanvb openismus com>]

On Mon, 2013-02-25 at 09:23 +0100, Milan Crha wrote:
> On Sun, 2013-02-24 at 23:01 -0500, Matthew Barnes wrote:
> > I went ahead and committed it for 3.8.
> >
> > We'll see if this has any impact on the GDBusConnection leak, but it's
> > an improvement regardless.
>
>       Hi,
> easy to test it with Tristan's [1], isn't it? You'll realize it'll not
> help, just the opposite, as I expect that creating a factory connection
> for each created book/calendar client only adds more things to be done
> on various layers, thus makes opening generally slower.

Milan, the factory connection is certainly something to consider here.

While you have a point that, we still wont be able to test it until
EBookClient/ECalClient can properly be finalized (i.e. the bug you
refer to regarding imbalanced reference counts needs also to be
fixed), we still need a way to release that factory connection.

Perhaps, if performance at connection time is very slow, we should
take an approach that the factory connection be released when
all EBookClients are finalized (with a sort of internal ref-counting
logic)... if of course, performance of opening the book is such
a bottleneck.

Cheers,
        -Tristan

> Just my untested opinion.
>
> The issue with [1] is also related to [2], if it'll help you to properly
> fix them both. If you do not like the already available tool for
> testing, then I can provide a simple test.c, which will exhibit the same
> issue - I only wanted to use something in place, to exhibit the
> regression from 3.6 code.
>       Bye,
>       Milan
>
> [1] https://bugzilla.gnome.org/show_bug.cgi?id=693461
> [2] https://bugzilla.gnome.org/show_bug.cgi?id=693464
>
> _______________________________________________
> evolution-hackers mailing list
> evolution-hackers gnome org
> To change your list options or unsubscribe, visit ...
> https://mail.gnome.org/mailman/listinfo/evolution-hackers