Plan:
8 Oct initrd builds start for rc2
21 Oct initrd builds finish (or earlier..)
last day for uploads of any udebs or debs for rc2
Reality:
12 Oct languagechooser changed to use UTF-8 for Hebrew
12 Oct start building "final" d-i initrds.
13 Oct build failed on hppa and m68k
14 Oct start building "final" d-i initrds, with fixes for hppa and m68k.
15 Oct finished building "final" d-i initrds.
16 Oct powerpc floppies overfull
17 Oct rootskel changed with no discussion to use utf-8 for serial console
17 Oct rootskel change tested (broke things) and reverted
18 Oct kbd-chooser changed to better support Turkish, Hebrew,
Lithuanian, Latvian
18 Oct languagechooser changed to install localization-config more often
18 Oct Turns out UTF-8 hebrew was untested. languagechooser change of
12 Oct reverted.
18 Oct rootskel finished building (after manual builds on some arches)
18 Oct ddetect changed to fix airport interface ordering
19 Oct mips initrds overfull. ths suggests that we switch mips to
partman. I have serious doubts. We remove stuff from mips
initrds that we know will be dropped no matter which route is
taken.
19 Oct kbd-chooser and ddetect finish building
19 Oct New sparc 2.6 kernel udebs uploaded, without any prior warning.
19 Oct Verification that ddetect change worked, no problems (it seems..)
19 Oct Turns out that the languagechooser locaization-config change
was not tested and that it now installs always, and that it
breaks on at least ia64. languagechoser change reverted.
As you can see, we have a lot of untested and poorly tested changes
being made, and this along with a general lack of testing of whether d-i
works on some architectures (despite the release of pre-rc2) has led to
the final initrd builds being pushed further and further back. It's now
11 days behind schedule, and we've exhausted all of our padding, so
every day we slip now will delay the final release. This means that any
given change that is made needs to be important enough to delay the
release for if it turns out not to work.
This is exactly what happened before the release of beta4[1].
This is quite like what happened before the release of rc1[2].
Can we please find a way not to do this? It's giving me ulcers.
--
see shy jo
[1] http://kitenet.net/~joey/blog/entry/beta4/
[2] http://kitenet.net/~joey/blog/entry/the_release_problem-2004-07-22-23-23
Attachment:
signature.asc
Description: Digital signature