On Tue, Feb 13, 2007 at 07:56:36AM +1100, Hamish Moffatt wrote:
> This is a two-way street though. Aurelien was trying to solve a problem
> he perceived to exist with the arm port. His solution has been rejected,
> but is the original problem being addressed?
] I am really upset by the way the ARM build daemons are managed. The
] packages are not uploaded regularly, with sometimes three days between
] two uploads. [...]
]
] All of that resulted in ARM being the slowest architecture to build
] packages. [...]
-- http://blog.aurel32.net/?p=33
I don't imagine Aurelien's any less upset, but as far as I can see, there
aren't actual problems with the way arm's keeping up at present:
http://buildd.debian.org/stats/graph2-quarter-big.png
The current "out of dates" according to britney are:
4 i386
13 amd64
25 sparc
32 arm
38 alpha
45 powerpc
47 mipsel
49 mips
55 s390
56 m68k
82 hppa
86 ia64
Which likewise seems to indicate arm isn't an issue.
As far as demonstrating the plausibility of setting up emulated buildds is
concerned, I don't think it makes any sense to do that by working on the
live archive for a release architecture. Personally, I've been trying to
promote emulated buildds since at least 2005, but you do that by diving
in yourself and producing a demo, not taking a release architecture with
you and having its users have to tread water with you if you turn out
to be wrong and have to find some way to undo it.
> Frankly I think ftp-master abused his dual roles (ftp-master and arm
> buildd admin) in this incident; any one else's actions would have been
> subject to peer review.
Uh, what's this if not peer review?
In addition, I reviewed both changes, and am not a buildd admin, though
I do share Steve's ability to do give-backs.
Cheers,
aj
Attachment:
signature.asc
Description: Digital signature