Re: [damien.doligez@inria.fr: [Caml-list] announcement: Objective Caml 3.08.3]
merge 279030 299722
thanks
On Fri, Mar 18, 2005 at 06:30:33AM +0100, Sven Luther wrote:
> On Thu, Mar 17, 2005 at 10:23:17PM +0100, Sylvain LE GALL wrote:
> > Hello,
> > On Thu, Mar 17, 2005 at 10:09:27PM +0100, Stefano Zacchiroli wrote:
> > > On Thu, Mar 17, 2005 at 08:46:41PM +0100, Sylvain LE GALL wrote:
> > > > If you made a debian repository on alioth, i propose to rebuild ocaml
> > > > package for arch i386 ( it was a big problem for me last time, because
> > > > i386 wasn't build in unstable, because of latency with lablgtk ).
> > >
> > > Well, this is the same I proposed with the exception I propose using
> > > experimental instead of an alioth repository.
> > >
> > > I think experimental is more suitable since it is more "official" than
> > > an alioth repository.
> > >
> >
> > Humm, well i don't disagree to your statement. There is only one problem
> > : do i have the right to upload to experimental ( considering i am not a
> > DD ) ?
> >
> > ( if i can, i agree on the fact that an alioth repo is not mandatory ).
>
> Well, we had the following reply from vorlon :
>
> 20:33 < vorlon> aba: I told him last time he asked, that ocaml is
> self-contained, so as long as he doesn't have anything broken in testing
> (which he currently does, I think -- I just saw an RC bug on mldonkey),
> it doesn't matter to me.
>
> So, this basically is a green light, provided the RC bug against mldonkey is
> fixed. Mmm, there are two of them
>
> Sylvain, can you comment on this ?
Sylvain, i feel bad about this whole mldonkey mess, we should have noticed
earlier and helped you out on this, we do quite a bad job as sponsors or
mentors or whatever on this account.
I have (quickly) read both RC bug reports, and it is quite murky. I am not
fully sure of how things are, but i would say :
1) chose a mldonkey group and user, and have it created and make sure it
works with it. If somebody has already a mldonkey user, output a note to the
user telling him to remove it or whatever. (or a more helpfull selection as
posted in the bug report).
2) make sure all files get created. /var/cache/mldonkey would maybe be a
better default than /var/lib/mldonkey.
3) if removing the file as conffile does the job, then let's do it. more to
this below.
4) take advantage of debconf's priority levels to ask at high (the default)
priority only the minimum set of questions, and go into more details at
level medium or low.
as for ucf/conffiles issue, i am not sure.
Could you patch mldonkey to look at /usr/share/mldonkey for its config file as
well as /etc/mldonkey ? Making /etc/mldonkey the default, maybe overriding
individual values from the share version ?
This way you would ship a commented out /etc/mldonkey config file as conffile,
have a big warning about the ucf handling, and that users should really use
dpkg-reconfigure to handle the file, but still give them power to override
this.
Let's fix this as soon as possible, and let's go to #debian-ocaml-maint this
evening if needed to work those issues out if possible.
A quick note to the reporters of this bug, well as i read this bug report the
first time, i found that the first reporting of the bugs where a wee bit too
agressive, especially coming from a debian developer. But let's hope we can
fix things in a fashion agreable to all now.
Friendly,
Sven Luther
Reply to: