* Marc Haber

> In the time of the freeze, a release critical bug is actually good
> since it might allow a new version into sarge ;)

  Well, that's a point, however I don't believe the release mangers
 would accept a new package in order to fix this bug, regardless of the
 severity.  Even if the bug was RC because of Policy, it doesn't affect
 Sarge because it only happens during upgrades from packages that was
 never in Woody.

  Besides, I want really, _really_, REALLY good reasons for pestering
 the release managers now, as I strongly feel that the one thing our
 users need right now is a new stable release, regardless of how buggy
 Munin in it.  So I'd rather not join in the avalanche of "please let my
 package into testing" begging choir on debian-release@ unless I feel
 it's absolutely necessary.  This bug doesn't even come close to being
 important enough, as I see it.  If I can speed up the release by two
 minutes by avoiding using two minutes of Mr. Langasek's time, that
 would be the best thing I can do for our users.  Because after all,
 Munin's a pretty insignificant package when looking at the entire
 distribution.

  Seeing your smiley, I assume you agree.  :-) 

> My suggestion would be:
> 
> (1) detect whether sysv-rc or file-rc being used
> (s2) read the symlinks and change them if necessary
> (f2) emit a warning, update-rc.d remove, update-rc.d insert
> 
> (s2) is the same as we have today, (f2) is better than failing.

  Hmm, but that'd make the file-rc method unconditionally nuke any user-
 customised startup ordering, won't it?  I'd rather learn file-rc (never
 used it) and see if it's possible to figure out if the user has changed
 the default ordering before doing the remove-insert thing.  It seems
 that sysvinit and file-rc is the only packages providing update-rc.d so
 I guess that means they're the only ones providing any init system at
 all.

  However, this bug will probably be "fixed" by the removal of the
 function in question after Sarge has been released - it won't serve any
 purpose then anyway.

> I have written e-mail to the maintainers of sysv-rc and file-rc
> whether they would be willing to provide read-rc.d

  That would be nice to have, yes.  Even better, something like
 "update-rc.d --update blah blah" which would only update the order if
 it wasn't changed by the user, or it could even pop up a dpkg-style
 conffile dialog asking the user if he'd want the new config as shipped
 in the package.  Possibly.

-- 
Tore Anderson



-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Reply via email to