
On Mon, 06 Nov 2006 23:23:23 +0100, Michael Prokop wrote:
... So I thought if it is possible to splitting the squashfs image into two images that are merged by unionfs, where the smaller one contains a basic system that includes all stuff from grml-small (which may be larger than 50MB, but as small as possible) and have a cheatcode to load only the small part into ram (and not use the large part at all). I do not know if this is feasible in the grml build process (obviously it will require some extra work of splitting the image), but it would be cool. I don't know any live cd that tries to do things like this, so perhaps it is not possible...?
Interesting idea. I'll add it to the todolist and will discuss it with other developers.
Thanks for the consideration. I wanted that long time ago. It is totally feasible, and has been done in other live cd system, like slax (http://www.slax.org/features.php):
,----- | Other Live CDs contain all software in a single compressed file. If you run | such a Live OS from CD-ROM, the CD drive has to seek back and forth really | frequently, because different files are located on different locations of | the CD medium. This makes the system notably slow. | | With SLAX, all conformable parts of the filesystem are compressed to a | standalone file, which doesn't contain anything else. For example, all files | which belongs to Xwindow are packed in xwindow.mo, KOffice related stuff is | in koffice.mo, etc. If you work with KOffice, you usually need only files | from KOffice and nothing else, and hence all files from that part of the | filesystem are separated from the rest of it, your CD drive has to seek only | in a 10 MB area. This significantly improves the speed. `-----
This is just an example. I think the most advanced Live CD regarding using unionFS is Puppy Linux, it allows 5 layers to work together, and it has detailed docs on how it is done:
http://www.puppyos.com/development/howpuppyworks.html
BTW, I have planned long time ago to write the author of Puppy Linux a personal email to invite him join on board to develop grml, but haven't go time to do that yet. :-)