Bought myself a USB DVD burner on Saturday. Trying to get it working on Sunday I learned that we are not 100% there yet in terms of DVD burning and handling. There where multiple things that did anything but just work. The first issue was that HAL and gnome-device-manager was rather flakey when it comes to handling the device. For instance if I booted my laptop with it connected nautilus etc., seemed unaware of its existence. Also if I plugged it in while having a blank DVD already in then my system also failed to deal with it. Only if I waited until the desktop was running, plugged the USB dvd writer in with an empty slot, then put in a empty DVD, only then did a ‘DVD+R blank disc’ icon appear on my desktop.

Trying to burn a myself DVD Nautilus complained that I needed to put in a supported media for example DVD+R disc. Problem was that a DVD+R disc was exactly what I had in the burner. Turned out that was a bogus error message and the real issue was lacking diskspace to store the temporary ISO image.

At this point I tried switching to GNOME Baker instead to create my DVD’s. At this point I also realized that handling two burners on the system was not something that worked two well. During my period of testing (which included a couple fo reboots) only my internal cdwriter was the only one recognized, only the USB dvd writer being recognized or in a few rare cases both where. That said even in the best of cases there seemed to be some permission issues when trying to burn as a normal user onto the usb dvd burner. In the end I tried running GNOME Baker as root, which did sorta work until I got error messages which I think where from cdrecord. At this point I decided to try out K3B hoping that since its been around a little longer it might have some workarounds or something for my issues. K3B seemed unable to handle the two drives also, but removing my internal writer physically eventually solved that problem. K3B did actually have a workaround or rather it didn’t use cdrecord, dvdrecord but growisofs to create DVD+R’s and some googling seemed to indicate that growisofs is the only linux library capable of doing so.

So to sum up there is still some work needed in the detection area and probably some library side work as well. At least ideally I guess we would want one intergrated set of libraries handling all kinds of cd and dvd burning. I will try to follow up on my problems by filing more bug reports, just need to figure out against what to file them, where to file them and how to find the information needed to file them first :)

As a sidenote I also tried Nero Linux. Since my burner supports ‘lightscribe‘ I figured the Linux version of Nero might have that as it was Nero for Windows that came bundled with the USB writer.
Discovered it didn’t and I also discovered that FC5 no longer is able to provide GTK1 based applications with any fonts it seems. Anyway know if there is any work being done on Linux ‘lightscribe’ support?

There was some complaints not long ago on Planet GNOME about projects using Werror as default for their compile and the value/cost of it. I am currently using latest Fedora Development which comes with an updated GCC version. This is causing the GStreamer build to fail from CVS (and we haven’t gotten around to fixing all the issues yet), but it has actually also shown why having it is a good idea. Cause atm GStreamer is the only multimedia project I am able to compile at all (by turning off werror) using GCC 4.1. Xine, mplayer and VLC are all failing to build for me. So by fixing all these things for previous versions of GCC and other compilers made GStreamer compileable with this latest incarnation. My point is that unless werror is turned on by default nobody is going to bother fixing those warnings, which means that the code is less likely to be compileable as is with other compilers or future versions of your current compiler. Also since the GStreamer community has done this for a long while now, incrementally fixing outr build with new versions of GCC etc., the job of removing the warnings isn’t overwhelming. I think that for some GNOME modules (and most software in general) the job of making them warning free has at this point become so big that nobody can manage to build up the energy to even try.

vmware server

We are planning on supporting Solaris with our plugins. Thanks to Sun we now have a SPARC box in our office to build and test our plugins on for the SPARC platform. Also to have a Solaris Intel build available I looked into the recently announced free vmware server product. Struggled a little to get it going on my Fedora development system as the config problem had some problems finding/using the kernel headers, but a patch called vmware-any-any-update97 sorted me out. One thing I have to say about vmware is that is it a very nice product. When using it you get the feeling of using something very polished and the GUI feels (and is) something actually made for running on your linux system, not an afterthough or sidejob of a GUI meant for a windows system.

GStreamer and GTK on embedded systems

I mentioned earlier some of the things that happened at 3GSM in
regards to GStreamer and GTK+. By combining that information with our current engagements at Fluendo and the dealings we have had at Fluendo in wake of 3GSM I think we are now in a situation where almost every major chipmaker in the embedded sector plans to or is already offering a SDK for their boards and chips using GStreamer for the multimedia handling and GTK+ for the GUI. And what is maybe even more interesting is that these efforts have come about due to the customers of these companies asking not only for a ‘linux SDK’, but for an SDK including GStreamer and GTK+. I don’t know the exact reasons for why our mindshare suddenly have exploded, but apart from the endorsement of GNOME and GTK+ from the major distributions I think we owe a lot of this to Nokia and the Maemo environment. Currently I know of hardware products under development using GStreamer and GTK+ such as TV’s, PVR’s, Media Center solutions, set-top boxes, portable music players, web tablets and mobile phones. It will be a year or two before these products starts hitting the market, but the current development is nothing short of amazing. The chipmakers seems believe that for embedded market will be focus on two major stacks going forward, WinCE and the Linux/GStreamer/GTK+ combo. Pretty cool stuff in my book :)

The effort to create a more beautiful effort is moving us quickly into a situation where the future desktop depends on hardware acceleration provided from the graphics card. The problem is that while 2D drivers are available as free software for the most part, the major graphics cards makers keep their 3D drivers closed. If you wanted good 3D acceleration under Linux the non-free NVidia and ATI drivers has been the way to go for a long time now.

So adding a ‘hard’ dependency on 3d hardware acceleration will implicitly mean we are basing our free desktop on non-free drivers feels like a step backwards. On the other side I don’t feel not going down the route of making our solution more visually appealing and hopefully also more usable is a viable option either, so I am not suggesting that these efforts gets rejected or stopped.

Hopefully the increased use of 3D will also lead to increased interest among free software developers to actually make free 3D drivers (and thus finding enough resources to overcome the difficulty of reverse engineering current cards) or that the increased use of 3d will increase the pressure from distribution makers and consumers to get ATI and NVidia to open their drivers. Anyway I hope I am wrong to be a bit worried about how this ends up and instead this leads to a big increase in contributions to things like MesaGL