This week in Anaconda

Posted on November 12, 2010 by Chris Lumens in work.

This week’s entry covers variable substitution in kickstart files, the new user interface, the default filesystem, and more.


Last week, I promised I’d talk about how substituting the $releasever and $basearch variables in repo lines played out. I’m sure many people have been waiting anxiously for the results. Well, this extremely involved commit from bcl fixed it all up. As I suspected, it was a problem where we had all the pieces but needed to hook them up right.


The user interface discussions are up and running thanks to mizmo’s thorough screenshotting and questioning as linked to from the UX Redesign wiki page. She’s raised quite a few questions about most every screen in anaconda. We’re going to be making a lot of decisions about the purpose of each screen and UI element in the next release or two. I’ve wanted this kind of discussion for years but something else has always preempted it. Now I think we finally have enough people working on anaconda to where we can afford to have someone go off and work on the UI. Just don’t expect instant changes.

Since we’ve just started, there’s little to talk about here. We haven’t decided or changed anything. I don’t even think we know what direction we should be moving in. Basically right after I finish this post, I’m going to go respond to everything on the wiki page to get us moving there. If you have thoughts, feel free to join in the conversation. However, I’d like to keep this focused on Fedora. Looking at other distributions is useful, but we need to be making decisions based upon what Fedora needs and where Fedora wants to go, and not just because some other distribution does something.


This subthread on fedora-devel-list is all about whether we’re moving to btrfs as the default filesystem. The short answer is no, not for F15. We will, however, allow you to create btrfs volumes without passing any secret command line parameters.

One reason for me not wanting to make btrfs the default is that I don’t yet know what all it’s going to be capable of doing. There’s been a lot of talk about it being a potential replacement for LVM, which makes me think we need a real strategy around it. What, exactly, do we want to do? Add to this that wwoods wants to quit using LVM by default anyway and it seems like we really need to come up with a plan. After all, I’d hate to kill LVM this release, switch to btrfs by default next release, and then move in some fancy btrfs LVM-like features for the release after that. That’s a whole lot of disruptive filesystem changing.

Who wants to talk about this at FUDCon in Tempe? I’d love to hear what people have to think about where we should be going.


As you may have read elsewhere, Will Woods is moving over to the anaconda side. I’d hate to speak for him before he’s even gotten a chance to start and look around, but it seems like he’ll still be working on testability and all that kind of stuff. It’s just that he’ll be looking at it from a different side and have commit access to make changes. So, welcome! Now he can be personally responsible too.


Another week, another major release. Once again, congrats are in order to the ridiculous number of fixes and features done by the installer guys. And just in case you were wondering:

~/src/anaconda$ git log -p rhel5-branch..rhel6-branch \| diffstat \| tail -1
 1390 files changed, 2364004 insertions(+), 2425494 deletions(-)