Re: revision control for the kernel (BitKeeper)

Larry McVoy (lm@bitmover.com)
Mon, 22 Feb 1999 00:24:00 -0800


Me:
: Size What
: .5M The 0.99.15 tarball, gzipped

dwguest:
: Hmm.
:
: -rw-r--r-- 1 aeb aeb 1230386 Feb 3 1994 linux-0.99.15.tar.gz
:
: .5M is a really rough estimate for 1230386 bytes.

Yup, sorry, my eyes are getting old, I was looking at

-rw-rw-r-- 1 lm wheel 1230404 Nov 26 16:18 linux-0.99.15.tar.gz
-rw-rw-r-- 1 lm wheel 545245 Nov 26 16:18 linux-0.99.2.tar.gz

and saw the second one, not the first one. So the numbers were not as bad
as I made them seem.

: A complete archive including lots of marginally related stuff
: takes 1.3 GB here, less than 10% of a $350 disk.

Point taken, but several people have pointed out that their resources are
limited. I'm very proud of how well Linux has done in very few resources
and I'd hate to be the cause of people saying "we don't support you anymore,
you are too slow/small/whatever".

And there was that thread recently about how 12MB for the kernel is too
much when it is uncompressed. Ted T'so was out here and we had a long
talk about it while out hiking, and he came up with a pretty cute idea:
you could hack /lib/cpp to know about BitKeeper (the revision control part
is written as a library so this isn't far fetched). With some caveats, you
could do builds from the compressed repository so that you never had to
unpack it - sort of like working inside a compressed tarball. Think about
it. It could be made to work, so that instead of the kernel source tree
getting bigger as a result of BitKeeper, it gets smaller... It's a nice
thought and if it ever happens, remember that Ted dreamed it up.

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/