At this point, it might be helpful to try and define the goals for the project a little more clearly... for example, what are we planning to deliver? Is the goal to create a generic build system for generating customized Fedora bootable CD's... or are we trying to assemble a "one size fits all" release with a flexible set of startup options?
How about a project website, cvs and download mirrors? Is this a good candidate for SourceForge or is Red Hat thinking about providing this support?
Last but not least, leadership. Is there a "benevolent dictator" at this point? :)
-Tom
On Sun, 2005-04-10 at 11:44 -0600, Tom Lisjac wrote:
At this point, it might be helpful to try and define the goals for the project a little more clearly... for example, what are we planning to deliver? Is the goal to create a generic build system for generating customized Fedora bootable CD's... or are we trying to assemble a "one size fits all" release with a flexible set of startup options?
Seen casper?
I think a generic "build system" for generating customized Fedora LiveCDs is what the aim of this project is. There is never going to be a "one size fits all"; thats not The Fedora Core goals either ;-)
How about a project website, cvs and download mirrors? Is this a good candidate for SourceForge or is Red Hat thinking about providing this support?
I think all this can happen in due time. The website, is something that can happen sooner than later (I don't mind being on task for that), cvs/download mirrors, well, more work needs to be done first
AFAIK, The Fedora Project will be providing support to this and it doesn't need to get hosted at SourceForge
Last but not least, leadership. Is there a "benevolent dictator" at this point? :)
No, aptly described, its anyone "that get's shit done"; while waiting, Greg is probably the default man for the job :)
On 4/10/05, Colin Charles byte@aeon.com.my wrote:
On Sun, 2005-04-10 at 11:44 -0600, Tom Lisjac wrote:
How about a project website, cvs and download mirrors?
... cvs/download mirrors, well, more work needs to be done first
From my experience, the work tends to get done when there's a place
for posting and sharing results. Since some of the largest components of this project currently exist, the release early, release often stuff could potentially start as soon as some basic architectural decisions are made.
Last but not least, leadership. Is there a "benevolent dictator" at this point? :)
No, aptly described, its anyone "that get's shit done";
That model seems to work pretty well when a project starts with a single idea or a blank page... but that's not the case here.
Neville has a complete (and excellent!) FC3 based CD and build system available right now. Steve has offered his Rookery Build system that looks very promising. I've also developed an architecture called EpikBuilder for creatiing CD, UML or disk images from rpm repositories:
http://theseus.sourceforge.net/index.php?wiki=EpikBuilder
... and I'll bet there are probably others on this list who have some great code and ideas to offer. The other GPL'ed live CD implementations are also available to borrow ideas and code from:
http://distrowatch.com/dwres.php?resource=cd
So IMHO, most of the hard work has already been done. What's needed now is a unifying spec or wishlist that can be used to pick, choose and merge the best parts of all this good stuff into a Fedora specific build system. I'd be willing to draft a strawman spec for you all to laugh at if one doesn't already exist. :)
-Tom
On Mon, 2005-04-11 at 02:19 -0600, Tom Lisjac wrote:
So IMHO, most of the hard work has already been done. What's needed now is a unifying spec or wishlist that can be used to pick, choose and merge the best parts of all this good stuff into a Fedora specific build system. I'd be willing to draft a strawman spec for you all to laugh at if one doesn't already exist. :)
It might be interesting if the people who have done live cds could list what changes they have done to make them work. Then we can take these changes and try to get them all into the distro. When everything needed for the live-cd is in the core distro, all that is required is a fedora-livecd-creator package with some script to create a live cd from a fedora core tree.
Some changes needed:
* squashfs This would be nice, zisofs is included in fc and works, but squashfs would perform better.
* init script changes There are likely a few init script changes needed. The stateless linux changes added some stuff to /etc/rc.sysinit that could be used. (The /etc/sysconfig/readonly-root stuff). Maybe some more changes are needed, we should then try to get them in.
* Additional packages Does the live cd need some package not currently in the distro. If so, we should try to get it in. (I guess this means both packages needed on the live-cd and packages needed to build it.)
What other changes do we need in order to make a script that can generate a live-cd from a random rawhide snapshot?
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= Alexander Larsson Red Hat, Inc alexl@redhat.com alla@lysator.liu.se He's an unconventional native American cat burglar with a passion for fast cars. She's a chain-smoking kleptomaniac safe cracker from aristocratic European stock. They fight crime!
Alexander Larsson wrote:
On Mon, 2005-04-11 at 02:19 -0600, Tom Lisjac wrote:
So IMHO, most of the hard work has already been done. What's needed now is a unifying spec or wishlist that can be used to pick, choose and merge the best parts of all this good stuff into a Fedora specific build system. I'd be willing to draft a strawman spec for you all to laugh at if one doesn't already exist. :)
It might be interesting if the people who have done live cds could list what changes they have done to make them work. Then we can take these changes and try to get them all into the distro. When everything needed for the live-cd is in the core distro, all that is required is a fedora-livecd-creator package with some script to create a live cd from a fedora core tree.
Some changes needed:
- squashfs
This would be nice, zisofs is included in fc and works, but squashfs would perform better.
- init script changes
There are likely a few init script changes needed. The stateless linux changes added some stuff to /etc/rc.sysinit that could be used. (The /etc/sysconfig/readonly-root stuff). Maybe some more changes are needed, we should then try to get them in.
- Additional packages
Does the live cd need some package not currently in the distro. If so, we should try to get it in. (I guess this means both packages needed on the live-cd and packages needed to build it.)
What other changes do we need in order to make a script that can generate a live-cd from a random rawhide snapshot?
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= Alexander Larsson Red Hat, Inc alexl@redhat.com alla@lysator.liu.se He's an unconventional native American cat burglar with a passion for fast cars. She's a chain-smoking kleptomaniac safe cracker from aristocratic European stock. They fight crime!
-- Fedora-livecd-list mailing list Fedora-livecd-list@redhat.com https://www.redhat.com/mailman/listinfo/fedora-livecd-list
Hello all, another two cents worth,
I think it would be wise to choose at least one compression technology and also use copy-on-write filesystems as this will allow maximum useability and least change to the installed distribution.
It would also be nice if the distro image generated for the CD was part of the Fedora installation processes, the workstation image is not suitable. A custom distro image should developed which includes some of the workstation and server set of RPMs and could be called the fedora demo or live cd image.
The Xendemo CD is a good starting point as this would the CD start multiple copies of Linux as well. The startup scripts are easy to understand and writing a new set of scripts for the Fedora livecd will not take long.
On Wed, 2005-04-13 at 10:02 +1000, neville wrote:
Hello all, another two cents worth,
I think it would be wise to choose at least one compression technology and also use copy-on-write filesystems as this will allow maximum useability and least change to the installed distribution.
It would also be nice if the distro image generated for the CD was part of the Fedora installation processes, the workstation image is not suitable. A custom distro image should developed which includes some of the workstation and server set of RPMs and could be called the fedora demo or live cd image.
The Xendemo CD is a good starting point as this would the CD start multiple copies of Linux as well. The startup scripts are easy to understand and writing a new set of scripts for the Fedora livecd will not take long.
This is an interesting idea. I know anaconda can run in a test mode where it just installs into a chroot. You could use that to have a super-easy graphical way to select what you want on your live cd. Like a graphical installer that installs onto a live-cd.
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= Alexander Larsson Red Hat, Inc alexl@redhat.com alla@lysator.liu.se He's a gun-slinging small-town werewolf on the hunt for the last specimen of a great and near-mythical creature. She's a sarcastic winged widow fleeing from a Satanic cult. They fight crime!
On Sun, 2005-04-10 at 11:44 -0600, Tom Lisjac wrote:
At this point, it might be helpful to try and define the goals for the project a little more clearly... for example, what are we planning to deliver? Is the goal to create a generic build system for generating customized Fedora bootable CD's... or are we trying to assemble a "one size fits all" release with a flexible set of startup options?
To me the most important thing is to be sure a live CD can be cut from rawhide, or any test release or final release (modulo bugs of course).
In other words the really important goal is to have a "convert this Fedora to Live CD" system rather than a one-off fork at a given point in time.
I personally think that for a variety of reasons all packages should be changed to work fine by default on a read-only filesystem, and detection of hardware should all be dynamic, so essentially the live CD changes should all be able to go "upstream" (into the standard packages) eventually. But doing that in the first cut probably isn't realistic.
Last but not least, leadership. Is there a "benevolent dictator" at this point? :)
Whoever does the work ;-)
Havoc
livecd@lists.fedoraproject.org