| 1 |
Hi, |
| 2 |
|
| 3 |
Here's why a good release cycle is needed: |
| 4 |
|
| 5 |
We have started QA on live environment and kernel of the 2004.1 release |
| 6 |
well in advance (even before the official releng date). I had |
| 7 |
everything ready to be uploaded a week ago. Livecd booted ok (released |
| 8 |
as _rc3 for testing). The only problem discovered was a missing /dev |
| 9 |
directory on the stages (catalyst bug). |
| 10 |
|
| 11 |
I did a cvs update to get the fix for this problem and rerolled the |
| 12 |
stages. Until here everything fine. I noticed I had forgotten vim on |
| 13 |
the older livecd, so I chrooted in it the stage1 and emerged vim |
| 14 |
manually. I reran the stage2 and uploaded the result, assuming the cd |
| 15 |
would be fine. |
| 16 |
|
| 17 |
Apparently the stage2 was modified by the cvs update too, and added a |
| 18 |
bunch of x86 rc-scripts to the default runlevel. These scripts called |
| 19 |
for the x86 only 'hwsetup' (which gave an ugly error), tried 'detecting |
| 20 |
the processor of the user' (which didn't work), tried loading APM bios |
| 21 |
stuff (ppc has no BIOS)... |
| 22 |
|
| 23 |
I am rebuilding the mini install and install cds (manually modified |
| 24 |
catalyst to del instead of add the rc-scripts), updated versions |
| 25 |
available today. Please note that in my REMARKS file in the catalyst |
| 26 |
main directory, I already stated months ago that it was not a good idea |
| 27 |
to assume all archs would use the same rc-scripts. |
| 28 |
|
| 29 |
Best regards, |
| 30 |
|
| 31 |
Pieter Van den Abeele |
| 32 |
|
| 33 |
|
| 34 |
-- |
| 35 |
gentoo-releng@g.o mailing list |
Archives