In the early days of NetworkManager I was not a fan. Fortunately it improved a lot and now it mostly does what I need (easier than anything else).
However there are a few things I don't like about NetworkManager and I hope the developers could look at these:
Tested on CentOS 5.2 with NetworkManager 0.7.0 from RHEL5.3. But same issues are reproducible with F10.
Also we have still some slots available for CentOS-related presentations, so if you have an interesting story to tell wrt. CentOS, or you know how to entertain a crowd, talk to Fabian Arrotin (or send a mail to the CentOS promotion mailinglist).
See you at FOSDEM 2009 at the CentOS booth !
When I installed CentOS 5.2 on my Thinkpad X200s only to discover that neither my onboard ethernet (e1000e), nor my onboard wireless (iwlagn) was supported by CentOS 5.2's 2.6.18 kernel, I managed to manually configure my Bluetooth UMTS connection to my cellphone to fix my connectivity problems and get back in business in no time.
Funny how the newest technology (UMTS) apparently is better supported by older kernels than ethernet or wireless (ieee 802.11) technologies.
My christmas present arrived early this year. Unfortunately, this one I had to buy myself but given the price-tag that is not a surprise.
I didn't need a new laptop per se, but the weight I was carrying sometimes (my own Thinkpad T43 and the customer Thinkpad T60) caused some neck pains. It probably is not completely related to that, but every excuse to buy a newer laptop is fine by me.
So what laptop was under the hypothetical christmas tree ? Well, the title gave it away already ;-)
This got me to think about something I remember discussing last LinuxTag during a Fedora dinner when Red Hat's Bugzilla was the subject. Currently RHbz is not indexed by Google and in my opinion this is hurting the Red Hat, Fedora, but also the CentOS community.
Today, in my quest for a media center solution that suits me, I started fixing some issues with running Elisa on CentOS 5. Elisa looks very interesting in the sense that it is written in python, it is a relative young project and it is esthetically pleasing (more than one of the other projects I investigated). And soon in RPMforge for CentOS (albeit not without issues yet).
I was not very impressed by the MythTV installation process (after having installed the packages for 0.21). The documentation does not really explain well what the different steps mean, but instead repeats the same labels as on screen with little or no additional information.
So if you get stuck because of something, the documentation does not help you at all. Lost a good hour because of doing the same things wrong over and over without a good explanation. And then because of sheer irritation simply added all the missing values directly in the MySQL database. Something was not right...
Browsing through the top500 supercomputers list, I noticed that in the OS listing, 5 supercomputers are running specifically CentOS (1%) while 389 are running some sort of Linux (not specified).
From the Linux list undoubtedly more are using CentOS, but the remarkable fact is that this known 1% CentOS is the same amount as the 5 Windows supercomputers.
The CentOS community is pretty limited in what we can do to the core distribution. Since our mantra is "aiming to be 100% compatible with Red Hat Enterprise Linux" we cannot fix bugs or improve the CentOS core without waiting for Red Hat to make those modifications first. We have limited leverage and a 6-month release cycle against us.
But that is not the complete truth, Red Hat usually has an internal, a vendor and a
publiccustomer beta period and everything that is found within that time-frame might get fixed before it is being shipped (and frozen) for the next 6 months.
Today RHEL 5.3 Beta was announced with a lot of interesting improvements.
I just released mrepo 0.8.6 with RHEL 4.7's RHN/up2date code included which makes mrepo work on other distributions without requiring to copy those libraries.
Some of the highlights include:
- Support for RHEL5 and CentOS-5.
- Added YaST Online Update support.
- Added fuseiso support (root access no longer needed).
- Added unionfs support to merge ISOs to a single tree.
- Faster relinking of repositories.
- Caching of directory indexes to prevent regenerating repositories.
The CentOS development team is looking into another solution for the CentOS website and forums. But there is no real knowledge or experience about Drupal (especially for forums).
So this is a request to the CentOS or Drupal community for people that have experience with Drupal for forums to join the discussion and help with the requirement and question to see whether Drupal would be an option for the CentOS Infrastructure team.
As we speak I am pushing the new wxGTK updates to the repository. It was needed in order to have a truecrypt package, but also required a lot of rebuilds and updates of packages that depended on wxGTK.
The good news is that this may bring us a bit closer to compatibility with EPEL, the bad thing is that the audacity builds fail (old and new versions) so for the time being no audacity, or no wxGTK update...
I also tried building the new VLC media player (0.9.2) but it had issues of its own so I did a rebuild of VLC 0.8.6i until I can fix it.
I have been updating my proposal for the new CentOS wiki frontpage. The aim is to reduce the number of links a new visitor sees, so it is easier to remember later or memorize the structure of the wiki.
A wiki is essentially unstructured and it requires determination and effort to put structure into a wiki so people can remember the structure subconsciously. That is the hard part with wikis and the part I dislike most.
I just released a new Dstat. I finally spend some time doing the boring release-dance:
- Verifying all changes since 0.6.7
- Backporting changes to python 1.5 version
- Creating the release archive without all pending patches and experimental stuff
- Verifying ChangeLog and documentation
- Testing on all Red Hat and CentOS/RHEL versions