Monday, June 12, 2006

some updates

What I did this week:

I wrote code that contains many "bashisms" but runs, this is the code that detects boot paramaters from the user for a few boot options. I would like to combine the boot params into on parser that sorts through all the user entered boot options instead of looping through them xxxxx times during the execution of casper-bottom scripts.

I had a rough time testing the changes I made to casper-ng, so I started to work on a tool that will be used to easily unpack, pack, and modify images created by live-package.
This did divert my attention for a while, but not too long.
  • Its called live-helper and can be found in the subversion repo.

Worked on documentation, explaining the theory of operation behind casper-ng.

  • I am currently in the process of finishing up the casper-ng documentation, you should be able to grab this in a few hours using subversion.

Cleared things up Marco, he will work on the initramfs hooks and I work on the casper-bottom scripts. I finished up removing some code that modifies the root files system at boot time and moved it to postinstall. However, I did go a little nuts with this, and was asked by Marco to keep this to a min.

  • I will work on casper-bottom scripts, he will work on the mkinitramfs hook (image detection and so on)

My plans for this week:

  • Adjusting casper-ng documentation after a peer review by the debian-live group - this will be ongoing for the rest of SoC.
  • Experiment with some hardware detection, creating some images and testing on as many machines as possible (I am limited to mostly i386) - plan on using discover
  • I think it will be possible to use yaboot to boot debian-live on ppc & ppc64 unfortunatly the only box I can test on right now is an experimental blade server that IBM donated to a research project I work on at school (not ideal)... I have to find some people who are willing to test ISOs for me or a good emulator.
  • Start examining the scripts executed once casper hands everything over to init on the rootfs - I get some weird errors somtimes, prob because there are some saved setting from the box that created the iso. Modules that are installed at boot time etc.
*POST AS MUCH AS POSSIBLE: my mentor is starting to freak out because I did a poor job of keeping him posted on my work.

Monday, June 05, 2006

My progress so far

I forgot to write a progress report for last week's progress...
Here are some things that I did after submitting an updated proposal:

  • Put a development box together and got it up and running w/ sid.
  • Identified many steps in casper-bottom scripts that can be added to the post install script of the casper-ng package and started to write that script. (Some deal with: kdm, gdm, xscreensaver, and creating the default user)
  • Started to map out the operations of a boot param script.
  • Tested a new way to create a user at creation or boot time w/o a set password.


Problems: Sid broke! - There is an error when trying to build a new iso from scratch.

Plans for this week:

  • Finish up documenting what I have already started to work on. This documentation will be located on the wiki.
  • Finishing up post install script for casper-ng.
  • Continue to work on init script to override defaults.
  • Start working on a post removal script to restore the rootfs to its original state.

Friday, June 02, 2006

New proposal and info

You can find a pdf of my new proposal here.

The latest source can be found at svn:\\live.debian.net\casper-ng

Visit the Debian-Live webpage for more info.

Note: I am pretty sure the page will be moving next week, so I'll post the new info once that site is up.

Friday, May 26, 2006

First Post

This is my first post since starting the project.

I am working with the live-debian project, but I think I will continue on with my original proposal.

Bascially that is to write an extention of debootstrap for creating live images.

Of cource this will work with the releases supported by Debootstrap includeing Ubuntu.

Right now I am looking at supporting a debian scheme which uses init scripts that closely resemble the INIT scripts from a normal debootstrap install. The other will take advantage of KNOPPIX technology, which is has been noted as having some the best hardware detection.

For now I will focus on researching the types of hardware detection.
This will probobly require me to create my own repository that contains both Kudzu (KNOPPIX) and Discover2 based hw detection.

There is a lot more to be looked into in terms of bootloaders, and types of read-only filesystems (cloop, squash file system, etc.)

There are no immediate plan to focus on incorperating any X-windows system or desktop into debootstrap-live.

Essesntially the core of this project is to identify the best tools for creating a working live version of the releases supported by debootstrap.