So! I made a Feature-styled wiki pagelet to keep track of the plans for the treebuilder branch of lorax:
https://fedoraproject.org/wiki/Anaconda/Features/Lorax-TreeBuilder
Build/test instructions will be updated soonish, but there's a link to my draft build instructions there. I'll also be writing documentation on the template language in the near future.
The important thing is this: I'd like to get us using treebuilder to build rawhide images ASAP - that means merging treebuilder into master. Is everyone OK with this?
And that leads to a second question: what do we do for F16? I don't feel like treebuilder is ready for F16 yet, but the squashfs/dracut-based implementation in lorax's f16-branch is.. immature, and very different from the treebuilder one. So, should we:
a) rush treebuilder for F16, b) keep lorax/anaconda images as-is for F16, even though we'll be changing it again in F17, or c) revert the squashfs stuff in f16-branch and build f15-style images again for F16?
-w
(note: I'll be on PTO until the end of the week, and Monday is Labor Day in the US, so I might not respond to this thread until Tuesday the 6th..)
On Wed, Aug 31, 2011 at 07:17:02PM -0400, Will Woods wrote:
So! I made a Feature-styled wiki pagelet to keep track of the plans for the treebuilder branch of lorax:
https://fedoraproject.org/wiki/Anaconda/Features/Lorax-TreeBuilder
Cool, I'll add livecd notes to this when I have the chance. And I still need to post my patches.
a) rush treebuilder for F16,
If it were ready, I'd say go for it...
b) keep lorax/anaconda images as-is for F16, even though we'll be changing it again in F17, or c) revert the squashfs stuff in f16-branch and build f15-style images again for F16?
If this makes the transition to treebuilder for F17 easier this is what I'd do. Assuming it doesn't mess up F16 Beta.
Given where we're at in the F16 process I think we need to make sure we don't break stuff :)
... I'll also be writing documentation on the template language in the near future.
What about debugging vs templates? When I got a traceback from a template in mid-August, the Python debugger could not plant a breakpoint there. The debugger would accept a request to break on a given file and line, but execution did not stop when that location was reached.
The important thing is this: I'd like to get us using treebuilder to build rawhide images ASAP - that means merging treebuilder into master. Is everyone OK with this?
Yeah, let's get this going on master right away.
And that leads to a second question: what do we do for F16? I don't feel like treebuilder is ready for F16 yet, but the squashfs/dracut-based implementation in lorax's f16-branch is.. immature, and very different from the treebuilder one. So, should we:
a) rush treebuilder for F16, b) keep lorax/anaconda images as-is for F16, even though we'll be changing it again in F17, or c) revert the squashfs stuff in f16-branch and build f15-style images again for F16?
Given where we are in the schedule, I'd say (b) is the best choice. (a) is too drastic, and (c) removes working (though maybe not perfect) code in favor of something we haven't tested in a while.
- Chris
anaconda-devel@lists.stg.fedoraproject.org