[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: quality of the cd images we are offering



Hello.

Since I have recently bought a CD-writer, and I'm also concerned about CD
quality and usefulness. I found this in CD #1:

home:/cdrom# find . -type f -name "kernel-source*" -printf "%f\n"
kernel-source-2.0.38_2.0.38-3.deb
kernel-source-2.0.36_2.0.36-3.deb
kernel-source-2.2.10_2.2.10-1.deb
kernel-source-2.2.12_2.2.12-4.deb
kernel-source-2.2.13_2.2.13-2.deb
kernel-source-2.2.15_2.2.15-3.deb
kernel-source-2.2.17_2.2.17pre6-1.deb
kernel-source-2.2.18pre21_2.2.18pre21-1.deb

This takes more than 92Mb from Binary-1 CD, which, in theory, was
supposed to be "as useful as possible for a single CD". We could use
most of that space for more useful packages than old kernel releases.

Should not we try to avoid putting so much crap in CD #1?

My suggestions are:

* If we ship different kernels for different architectures, put in the
first CD only two kernel sources: a) The one shipped with that
architecture and b) The last stable one, if it differs from a).
We could put the remaining ones in the last CD or just not include
them in CDs. If there is an absolute need that the CDs matches the
archive, we could move those old kernels to the "data" section we plan
to create, so they they are still available, but not part of the
archive we ship on CD.

In either case, we could really make an effort and release 2.2r3 in
sync (kernel-wise) for all six architectures, as we already do with
the testing distribution.

* If we still support Linux 2.0.x, we could package 2.0.39 and drop
both 2.0.36 and 2.0.38, or we could just not include those old kernels
in CDs (or move them to the data section).


Also, I wonder if it is really useful to waste 30 Mb of data for the
1.20 floppies almost nobody uses. Would be possible to write a simple
script which generates the base-*.bin files from base2_2.tgz etc. and ship
the script instead of the images?

Thanks.



Reply to: