The smolt device statistics page is working again and can be used to learn more about Fedora users. In the meantime, the feature list for Fedora 8 looks promising.
Smolt Statistics
Smolt is ging to hit the 80k mark anytime soon – not even three weeks after it passed the 50k mark. That’s a nice improvement. And the first OpenSuse machines have registered with Smolt at the stats page as well (and machines called “Lightspeed Planet 6.3”, never heard of that…). I’m looking forward to that development, it would be nice to see more distributions contributing to this database by default.
Also, the devices page works again as expected. And again the video devices give interesting results: while most Linux users should by Intel or at least Nvidia cards for the sake of good drivers, ATI/AMD cards are actually more spread:
But even more strange is that out of the 80k registered devices more than 8% (so roughly 6 500 devices) are not real but working in a VMWare.
I mean, sure, it can be expected that some people use VMWare to test the new Fedora or even to have a second instance to screw up or play around with, but more than 6k? That is a lot more than expected. I’m not really sure what to think of this.
Fedora 8 Feature List
The Fedora 8 Feature List is now out for quite some time and therefore worth a look.
First of all it is notable that Boot-Shutdown-Speedup idea is not listed at all. I wonder if this is an error or shows that such improvements will not be included the next release – again.
But about the positive stuff: of course for me most interesting is the inclusion of KDE 4 packages. If it works out Fedora 8 might be the first, larger distribution using KDE 4!
Also something I have my eye on is the plan to activate NetworkManager by default. This actually means to bring NetworkManager into a state where it is stable and reliable enough to be activated by default. This means using D-Bus instead of the current socket based way to talk to wpa_supplicant, providing system wide configurations and also support for multiple active devices.
Some of these features are already included in the current NetworkManger release and I do hope to see the new 0.7 release in Fedora 8. And I’m curious how this will work together with KDE4/Solid, btw.
Another thing I wasn’t really aware of is that nowadays a centralized font server as Xfs is not longer needed anymore. Therefore the Fedora 8 team stepped up with the aim of removing Xfs – Qt and Gtk use client side font rendering anyway.
I’m not sure about the technical details but if that description is right we might see speed improvements in the startup as well as some small improvements when the machine runs (because there is one process less).
The Laptop Improvements Plans are also interesting to me since I do use a Laptop – and had quite some Suspend problems (although these couldn’t be solved in the way the Laptop Improvements try to help).
In the way of system administration the new syslog program rsyslog might come in handy. Regexp based filtering of log messages as well as native database support and especially a realtime analysis framework for logmessages can be very helpful when you run a server for example.
However, there are also some things I’m a bit careful with. For example the codec helper. While the gstreamer based implementation should at least be no problem in KDE 4 (because there will be a Gstreamer-Phonon backend) I somehow doubt that the developers will even bother with KDE here. This is still Fedora…
And I also have a problem with another feature: the PulseAudio integration. This would mean to introduce a sound server. And I don’t want to have that.
We had a sound server in KDE with aRts. It was a great idea back then, but today it isn’t. Every additional stack of software introduces latency – and a sound server is used to introduce a lot. And in the days of youtube and VoIP you don’t want to have any latency at all.
Also, we have gstreamer or even xine as a sound backend – together with Alsa it can already handle all the important bits you might want to have today. As an example, KDE’s Phonon together with Xine can handle such things as muting music during a phone call just fine.
Of course I might be wrong here – I haven’t tried PulseAudio yet on any system. But before I try it I would like to know how they avoid the latencies you have usually by introducing such a heavy thing like a sound server. And I haven’t seen anything about that in the provided presentation (careful, 3 MB PDF).
In any case there are quite some interesting new features coming up with Fedora 8.
/home/liquidat 
I think in a lot of ways a feature of “faster startup” is one of those things that is doomed to not be fully successful. Instead, we should have discrete features (like the xfs one) where boot speedups are a good benefit and also just try to do less on boot. In the long run that will end up doing more to improve the startup experience.
For Phonon, I think we’ll be defaulting to Xine, at least for F8, because there is no completed GStreamer backend yet AFAIK (just 2 incomplete attempts). This means you will not get any of the Fluendo codec stuff at all in KDE apps. There isn’t much we can do about that. (I personally hate that feature anyway, but that’s irrelevant given the technical situation.)
As for PulseAudio, getting Phonon to output to that won’t be hard, we can just select the appropriate Xine output plugin. We can discuss that in the KDE SIG meetings, but my personal opinion is that we should follow what the rest of Fedora is going to default to there. Remember this is just about the defaults, setting the output in Phonon to something different (as a user) is trivial.
I have registered several CentOS systems using Smolt.. I can see them in the data base when looking up their databases.. but they do not seem to show up in the OS section..
Also why rsyslog and not syslog-ng?
Jeremy: At least in my tests parallel booting with for example initNG cut my boot dramatically – I had less than 35 seconds instead of 90 senconds.
Kevin: Of course we should follow the Fedora lead there. However, I still have to see the major benefits of PulseAudio. I had too many too bad experiences with aRts in regard of latency that I’m just very, very careful about such ideas.
Stephen: several entries in the OS section vanished some days ago – I have no idea why. There were also several different CentOS entries, but maybe they introduced a threshold, and your entries were of different types (like CentOS 4 and CentOS 4.1) and are therefore not summed up?
About rsyslog and syslog-ng: I never checked the advantages, so I don’t know. But you can find some discussions about that topic here.