Summary do D-I IRC meeting 20040619
The full log is available at:
Attending the meeting (sorry in advance for those of you I missed...):
fjp, Frans Pop
bubulle, Christian Perrier
madduck, Martin Krafft
migus, Pierre Machard
kmuto, Kenshi "sleepy" Muto
waldi, Bastian Blank
amck, Alastair McKinstry
eppesuig, Giuseppe Sacco
jekis (sorry forgot to note your real name)
p2-mate, Peter de Schrijver
tbm, Martin Michlmayr
luther, Sven Luther
ths, Thiemo Seufer
gotom, Masanori Goto
and probably a few more people lurking..:-)
Excused: Joey Hess, Steve Langasek, Colin Watson, Joshua Kwan and all
who cannot attend for various reasons of course...but will
During this meeting, we tried to get some progress and objectives for
the points identified currently as "release critical" or "bad" in a
recent summary by Joey Hess
We tried to assign some people "tasks"....these are sumarized at the
end of this mail.
1) Towards tc2 - RC bugs
A) some broken m68k images (unknown mke2fs fix needed)
No reaction on this point. Still opened. We need to find people for
fixing and testing
B) broken sparc64 module loading (fix: busybox-cvs 20040415-3)
NEEDS also netcfg to be pushed. Is it release quality?
Lot of discussion around this point : is netcfg release quality or
Indirectly, sparc64 module loading depends on this, so deeply testing
netcfg for *regression* with regards to tc1 is needed. tc1 netcfg was
considered satisfactory though not perfect.
-->Bastian Blank will test netcfg with PTP device
-->Martin Krafft will make regression comparisons between tc1 and
-->Giuseppe Sacco will test WiFi on PPC
C) messed up line drawing characters in base-config (fix: base-config 2.29)
CONFIRMED AS WORKING
Base-config has been pushed by Martin Michlmayr. Line drawing issues
Japanese with 2.6.6 needs close tests because of broken 2.6.6. During
the meeting, Korean also confirmed as broken in these circumstances.
-->Kenshi Muto follows 2.6+JP in sid_d-i
Pushing the unstable 2.6.6 kernel into testing is probably needed.
D) broken support for 100 gb partitions on i386 (fix: base-installer 0.083)
PUSHED by Martin
Martin Michlmayr point to the bug reporter : Dan C <firstname.lastname@example.org>
We need asking him to confirm if fixed in sid_d-i
Peter de Shrijver has the machine but cannot erase its disk...
E) the above problem seems to affect nearly any size partitions on alpha
and probably other 64 bit systems (workaround: base-installer 0.083;
also needs *-installer fixes)
WHICH -installer need to be corrected and pushed?
Someone needs to care about this and interact with 64bit systems
No candidate found..
2) Towards tc2 - Bad but non RC points
A) 2.6 lvm and raid segfaults and hangs (fix: lvm2 2.00.16-2)
Giuseppe Sacco volunteers for testing this on sid_d-i. Martin
Michlmayr mentions he has volunteers also.
If tests are OK, lvm2 needs to be pushed to testing.
B) airport module not available post-reboot on powerpc (fix: ddetect 0.101)
PUSHED but not confirmed working
Giuseppe Sacco will test while testing netcfg+wireless. If OK, ddetect
to be pushed to testing (needed also by Firewire)
C) firewire cd support (fix: ddetect 0.101)
PUSHED and confirmed working
Nothing more than pushing ddetect in testing (CD does not work in 2nd
stage but this is not d-i related)
D) mips Installs on r4k-ip22 needs 36 mb ram (fix: unknown -- new glibc?)
Thiemo Seufer will investigate this (seems to be close to be RC for mips)
E) broken ataraid and ida support (fix: libdebian-installer 0.26.really.0.22)
We need finding people with SATA things and have them test sid_d-i
F) 255077 : installs wrong kernel
Martin Krafft couldn't reproduce the problem. Will investigate with
3) Build problems
The netboot images do not build correctly on gluck currently, which
prevents real new builds for netinst/businesscard images.
We need to find someone who can ask for libdebian-installer4 upgrade
Martin Krafft will mail Colin Watson about this. Not sure if Colin can
4) Other points
Martin Michlmayr added doc/testers.txt listing posible testers for
various points. Please try updating this file.
Thiemo Seufer mentions the high number of packages non synchronized in
testin (different versions on different archs). He will post a summary
Bastian Blank continue working on ssh (bubulle: I don't remember what
it's needed for..:-))). There's some code but not working.
Discussion about new languagechooser. Several people push for getting
it in tc2 (a few minor non blocking points raised by Martin Krafft). I
mention that an earlier agreement with Joey at Debconf was avoiding
pushing it in the next release unless enough tested. The current
delays have helped in testing it properly, so maybe we can re-discuss
this with Joey.
Pushing languagechooser needs pushing countrychooser and lowmem at the
The confusion about image names is raised several times. It seems
that, except for "d-i old timers", the sarge_d-i, sid_d-i, daily
builds things are really confusing. The most important confusion is
"daily" which is indeed sarge_d-i while most users expect it to be
"cutting edge version".
I have shown a propoal I received from Vince McIntyre:
We agree that d-i page needs some rewrite about all this. We probably
need to hide tc1 now and focus people on sarge_d-i and sid_d-i. Also
discourage installation reports for beta4 when not further tested with
Summary of tasks
-netcfg with PTP device. Not RC for netcfg
-regression test for netcfg. If OK, we push netcfg and busybox-cvs to
-confirm or not 255077. Close if not confirmed
-ask Colin Watson if libdebian-installer4 can be upgraded on gluck
-test WiFi on PPC. Not RC for netcfg. Test for airport support
-test raid+lvm, coordinate with volunteers mentioned by Martin Michlmayr
-make a precise point about 2.6+CJK with regards of kernels currently
-ask "Dan C" about >100Gb partition bug reproducible or not
(DONE asked 6/20)
-with Matt Kraai, probably, try to have a clearer rewrite of d-i main
page about images
-make a summary of unsynced packages in testing
-investigate needed 36Mb on mips rk4-ip22
(not sure because of personal time constraints)
-coordinate raid+lvm tests (w/ G. Sacco) and push lvm2 in testing if OK
-see C. Perrier tasks...:-)