The basic idea is to use the virtio devices from the initial installation to avoid all the work and hassles involved with changing the drivers later. This avoids mistakes which can lead to an unbootable guest or “Local Area Connection 2″ annoyances. There are two big problems, however. Finding the latest drivers is not easy (unless you pay for RHEL, which might make it easier). I’m not even sure the ones used here really are the latest, but they seemed to be the latest I could fine. Second, the limitations of the virt-manager + libvirt + KVM stack mean you can’t connect both an installation CD (image) and a virtio driver CD. Only one virtual CD is allowed by some level in that stack; I’m not sure which one or why.

1. Download the virtio drivers. The copy I used was virtio-setup-200908.iso as linked from an article titled
   Preview of Redhat 5.4 Windows Virtio Drivers Part 1 from Haydn Solomon’s blog.
   wget http://linux-kvm.net/sites/default/files/virtio-setup-200908.iso
2. Download a copy of my convert-iso-to-floppy script which uses mkfs.msdos to create a floppy image and copies the files from the virtio .iso file using loopback mounts, which is explained by untitledfinale.wget http://coderich.net/downloads/convert-iso-to-floppy
3. Make the script executable.chmod +x convert-iso-to-floppy
4. Run the script. ./convert-iso-to-floppy virtio-setup-200908.iso
5. Create a new virtual machine as you normally would for a Windows guest, stopping just before clicking Finish.
6. If you have a new enough version of virt-manager, you can check the “Customize configuration before install” checkbox. Otherwise, click Finish, stop the virtual machine, open the details, reconnect the CD (or other installation media) if necessary, and reconfigure the guest to boot off the CD (or other installation media).
7. Delete the IDE disk device and re-add the storage as a Virtio disk.
8. Change the NIC’s device model to virtio.
9. Add the image that convert-iso-to-floppy created (which will be named the same as the iso, but with an .img extension) as a virtual floppy device.
10. Set the floppy to be readonly for safety. (This is optional, but it makes me feel better.)
11. Switch to the console view, run the virtual machine, and start the installation as normal.
12. When the disk configuration step comes up, no disk will be detected. This is normal.
13. Click Load driver…
14. Windows will automatically search the virtual floppy drive and find the drivers.
15. Select the virtio block storage and virtio network drivers for the Windows version being installed, using the control key to select multiple items as always.
16. Click Next.
17. Proceed with the installation as normal.
18. Delete (or disconnect) the floppy device, as desired.
19. Reconfigure the boot options, if necessary, as desired.
20. Disconnect the CD device, if necessary and desired.

I’m not sure why…other than Mark Shuttleworth seems to love copying Apple.

However, this change made me realize one thing… I never use the menu from the title bar. I use it on the task bar a lot (for “Close”, “Always On Top”, “Move to Another Workspace”, and “Always on Visible Workspace”, in that order of frequency), but not from the title bar. This is a random observation, nothing more.

I understand his push for “GNU/Linux” a lot more now. It’s one thing to talk about credit–there I disagree, because I don’t want to call it “GNU/Linux/X/GNOME/OpenOffice.org”. However, he made the point that calling the distro “Linux” points all the attention towards Linus, who isn’t pushing software freedom. I hadn’t thought about that before.

On a side note, RMS pronounces it “GNU slash Linux” or “GNU plus Linux”, where I think “GNU Linux” is a totally more reasonable… After all, nobody pronounces the slash in “and/or”.

Update: Apparently gnome-blog doesn’t set the title properly with BloggerAPI. I switched to MetaWeblog in the options and now it works!

You can find the article here: A First Look at Gaim 2.0 [linux.com]

After several attempts with the linux.com comment interface (and three mispostings), I finally created this posting: Various Thoughts

The Slashdot article is here: A First Look at Gaim 2.0 [slashdot.org]

In general, when Gaim hits Slashdot, lots of users spend lots of time making lots of comments. Many of these comments represent honest questions, many are completely irritating, and a few can be very insightful.

I try to address the legitimate questions as much as I can, because I like being helpful.

However, when it comes to the second category of comments, some things just really drive me insane. Here’s a good example: Re: What happened to Gaim-vv I don’t disagree with them that having videoconferencing support would be nice. However, to suggest that the developers just “hop to it” is incredibly irritating. (I’d love to write voice and video support, but other things are more important to me, and I have limited time.) It’s not just this one individual either… this is a systemic problem with large numbers of open-source users.

As far as insightful comments go, I really liked this one: Agreed, but people need to know that.

I’m biased, but I think my reply to that comment, and especially my conclusion paragraph therein, applies nicely to both of these situations. I’ll reproduce it here:

The whole issue of “do it yourself” is very complicated. In many cases, people ask for huge features (say voice/video support) and expect that it’s incredibly simple to do. Also, users often think they should still be treated like “customers” when they’re not contributing anything to the project. It’s not fair to take something for free and then complain that you want it improved to do what you want, but you don’t want to help in any way.

The amount of work required to fix a bug or add a new feature is highly dependent on the size of the change, probably even more so than the size of the whole codebase. The Gaim software is pretty compartmentalized. When I started out, I fixed a few bugs without knowing much at all about the general architecture. So, it’s not really that hard to fix little things. The problem is, *most* of the little things have been fixed already. Addressing more fundamental issues does take quite a bit of startup time to learn the codebase.

In general, I try to help out anyone who has a legitimate bug report, suggestion, etc. However, the amount of things that the developers can take on is extremely limited. There are a million things on my TODO list, and that’s just the top items I’d like to accomplish. So often times, “do it yourself” is the only response.

There is definitely a big opportunity for non-coders to help, though. Graphic design, web design, bug triaging, documentation, etc. are useful to all open source projects.

People who like open source because the product is free need to remember that the product is free. If they expect anything more than zero support, they’re going to be disappointed. People that view open source as a community are much better off. Not everyone in the community is a coder, but everyone that contributes is important.

– Originally posted to Slashdot as comment #16585466.

I’m not sure what the solution is. Maybe I should just ignore these people. Maybe we, as an open-source community, need to better communicate with our users, encouraging more non-code contributions?