[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: ocaml 3.08.3 upload plan (we are going to sid for this).



On Fri, Mar 18, 2005 at 10:27:21PM +0100, Sylvain LE GALL wrote:
> Hello,
> 
> On Fri, Mar 18, 2005 at 08:28:23AM +0100, Sven Luther wrote:
> > On Fri, Mar 18, 2005 at 08:15:18AM +0100, Sven Luther wrote:
> > > I ask you also to quickly look at your packages, and report here any RC bug
> > > that you may still have open, and what the actions for sarge are in this
> > > issue.
> > 
> > Listing of the March 11 RC bug list include : 
> > 
> > Package: coq-doc (debian/main)
> > Maintainer: Samuel Mimram <samuel.mimram@ens-lyon.org>
> >   294865 [        ] [X] coq-doc: The documentation for coq is non-free (this
> >                         issue is being solved)
> > 
> > Package: mldonkey-server (debian/main)
> > Maintainer: Sylvain Le Gall <sylvain.le-gall@polytechnique.org>
> >   279030 [        ] [X] mldonkey-server: debconf configuration is entirely
> >                         ignored
> > 
> > None of those really matter for this plan, so should not be an issue.
> > 
> 
> As promise, i can answer a little bit longer on the reason why mldonkey
> is RC for now.

Thanks.

> Well, the first point is that mldonkey doesn't have made release for more
> than 4 months now. A new release ( 2.5.28.1 or 2.5.30 ) is available
> since one week. The former version doesn't work very well since many p2p
> network has made incompatible release of their protocol in order to make non
> commercial client unable to communicate ( as usual they don't want
> people not install spyware/adware and use their network ). 

Ok. But what about the not edonkey protocols supported ? 

> Due to this perpetual evolution of protocol, i am not sure mldonkey is
> well suited for being in a stable release ( i think a mldonkey package
> in sarge will be useless after 4 months of existence ). So i am not
> sure, i want to spend my time answering mail concerning the fact that
> mldonkey doesn't work anymore... 

Maybe this would be a good candidate for the volatile distribution thingy they
have been speaking about. Can you investigate this ? 

> Last, but not least, mldonkey actual package layout "sucks" ( to quote
> one user that just send me an email ). In fact, i agree on this point,
> since 1 year i have try to conceil policy, usability and easy
> configuration of the package.... The result is a pretty big mess. It is
> not really a question about bad maintenance or bad sponsorship. I really
> don't think so. It is mainly the result of too much layer of compromise
> ( compromis en francais ). 

Let's solve this together, ok, i believe you simply lack the experience to
know how to best deal with this. An install at high priority should just work
and ask the less possible questions. You can do more advanced configurations
at medium or low priority.

> I have plan concerning the future upload, i need time to apply them : 
> - split mldonkey-server to extract the init script and configuration.
>   This will allow user to install mldonkey-server and not run it (
>   because half of the user use it only from time to time, not as a
>   daemon ). The name of the new package will be mldonkey-server-daemon. The
>   former debconf question concerning the fact that mldonkey should start at
>   boot, will not be asked anymore ( but a variable will stay in the
>   configuration file ).

Mmm, ...

I think it is best to ask as a high/medium debconf question if you want to run
it automatically, or manually.

> - reduce the number of question asked in debconf : i will reduce the set
>   of question to 3 or 4 and only for options concerning things that is
>   needed to create a base download.ini. It is mandatory, since without
>   initial download.ini the things is totally unsecure. I will integrate 
>   a sane configuration for the file /etc/default/mldonkey.

I think you should do that for high priority, but ask more stuff for lower
priorities.

> - ask for the create of a user mldonkey or p2p or debian-mldonkey.

Just create a mldonkey user and be gone with that. Maybe make the user
configurable in lower priority, and complain loudly if the user creation fails
because the user created it previously or something.

high priority installs should just work with the minimum of user intervention.

> For now, i am finishing a library ( ocaml-gettext ). Since i have nearly
> finish my work on this, i will end it before working again on mldonkey (
> i think that it should be ok by the end of the WE ).

Cool.

> Anyway, the package is no more in sarge, so the bug is tagged "sarge
> ignore". That should be no problem concerning the migration of ocaml
> 3.08.3 in sarge.

Yep, thanks.

Friendly,

Sven Luther



Reply to: