Sunday, July 25, 2010

(GSoC) And god said: "Let there be an Internet Gateway"

Hi, all.

Last week I said to you that HUPnP did not recognize UPnP internet gateway devices in the network. Now I think I can say it used to not recognize. :-)

Seems like Tuomo's patch fixed the problem and now the Solid UPnP backend is able to "see" my wireless router and finally invoke the UPnP actions available on it (like to do some port redirect and toggle the internet access, for instance). I have made some tests and seems like the action invocations are working well from the HControlPoint instance in the UPnPDeviceManager. Now it's time to refine the API and add more operations, probably.

However, "the world is cruel". This week Nikhil and Bart pointed to me and Kevin some problems with the media server stuff. Besides some mistakes on the HUPnP use from my part, seems like there is something wrong with the creation of the UPnPDeviceManager. According to him, due to the use of a QThreadStorage on the backends manager, a new HControlPoint is created for every thread using Solid, causing some inconsistent results. The expected behavior is only one control point across Solid stack (frontend -> iface -> backend). I'm working on the backend part, but the Solid thread-affinity stuff I prefer to let the more experts metalworkers to decide what's better.

Another thing Nikhil opened my eyes is about that I'm not running KDE trunk (I'm using that script development strategy from Techbase). I know, it's a blasphemy and I will build and use the SVN trunk ASAP. My unit tests are not enough and I need to do real system tests.

Well, I really hope this week to get the backend working fine to go into the little desktop integrations and the debugging app using an UPnP framework developed by some fellows here in the lab, called BRisa. I have to run a little bit with the schedule.

See ya.

No comments:

Post a Comment