Geoclue and WebKit: unit tests

[Javier Fernandez Garcia-Boente]

On Thu, 2010-06-24 at 23:56 +0300, Stefan Kost wrote:
> Am 24.06.2010 20:44, schrieb Javier Fernandez Garcia-Boente:
> Thanks for the patch - just some small patch review:
> 

Thanks.
I'll add thos changes to my git repository. I have to admit that it was
a patch not very detailed and overall, too much focused on what the
WebKit unit tests was requesting. I think this patch should be
implemented in a more general way, more aligned with what GeoClue wants
to provide to the users. However, I wanted to start first a brief
discussion about what is the best approach to proceed to fulfill the
requirements of the WebKitGtk port Geolocation unit tests.

First of all, as i already commented before, I think that GeoClue be the
first implementation of the Geolocation API of the WebKit project is
something very interesting for GeoClue; WebKit is gaining more
importance nowadays and probably not too much effort should be invested
to, at least, complete the implementation of the unit tests.

The best approach in my opinion would be to implement the Position
interface on the Manual provider, which is the one best fitting, from
the conceptual point of view, with the unit testing purposes. The
Localnet provider could be also a good candidate; i've already been told
there is already a bug and a patch proposal for implementing the
Position iface on the Localnet provider.

Another question would be if the SetPosition method, the one used for
setting a dummy coordinates, should be exposed by the Position
interface. I dont see too many uses of such method, besides unit
testing, so i would not go for it. 

However, this way requires the weird constructor added in the
GeocluePosition class, geoclue_position_iface_new. The problem is that
the MasterClient should be used to set the dummy location if we want to
be aware of the position-changed events. even though we only need them
because the Manual provider has the "ProvidesUpdates" flag.

So, the goal of implementing all the WebKitGtk Geolocation unit tests
would require a deep analysis of the current Geoclue design and perhaps
some internal changes, which might be i haven't enough knowledge about
to analyze or propose new approaches. I would like to know the opinion
of the ones with more experience and knowledge about the Geoclue design
and internal structures. 

Anyway, thanks again for your comments and they will be added to my
local repo.

Greetings,



-- 
Javi
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: This is a digitally signed message part
URL: <http://lists.freedesktop.org/archives/geoclue/attachments/20100625/2fb040e6/attachment.pgp>