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

Re: Communication on Debian Science Was: (Re: New Debian Science metapackages (Was: debian-science_0.16_amd64.changes ACCEPTED into unstable))

On Wed, Apr 25, 2012 at 12:07:56PM +0200, Michael Banck wrote:
> I certainly think this is a great goal, but I have some reservations
> about the process.  You are using debian/upstream, which in no way is in
> some science-related namespace and appears to be for general-purpose
> upstream metadata.

I admit that the current approach is a bit bottom-up.  However, compared
to having "nothing" in the package space this is some extra benefit.  We
started with the (also not longish discussed) Publication-* fields in
the tasks files which turned out as not flexible enough.  When working
with the > 100 Reference entries in debian/upstream files I regard this
way as pretty useful.  We could also gather similar information like
"Registration", "Donation", "Webservice" etc. - just have a general file
to dump structured information into seems pretty reasonable to me and
I see no reason to use some "science-related namespace" (which we also
do not for any other metadata file - so why should we in this case).

> Therefore, I think before wide adoption, this should
> be discussed on debian-devel and/or be a DEP.

I agree that a DEP seems to be some appropriate means to discuss this.

> Presenting this as a fait accompli to -devel after x packages have been
> converted might result in some irritation or confusion.

What do you mean with "converted"?  Converting from what?  Did I missed
some other means to specify references?

> Possibly I have just missed that conversation though, so excuse me in
> that case.

The problem is that nobody is really willing to discuss those issues and
most people simply throw ENOTIME.  IMHO we just *should* take time for
this but we all are free to decide where we will spend our time.
Charles is fighting since years for establishing debian/upstream but I
think that only a hand full of people realised this and finally support
this.  At DebConf 10 I also brought this into the Debian Science
discussion which is documented in the Wiki[1].  I warmed this up in last
years DebConf in the Debian Science BOF and sometimes here on the list
(I guess if you seek for

  site:lists.debian.org/debian-sciende "http://wiki.debian.org/DebianScience/ProblemsToWorkOn";

you will get the relevant threads because I was always linking to this.)

I have drawn the conclusion that discussion leads to plainly nothing and
thus followed the Do-O-Cracy principle to just do something.  I will
join the discussion about this once there is a counter proposal which
solves the problem better than debian/upstream.  I also do not see a
problem in converting the structured data from debian/upstream if
something else might win in the discussion - and I volunteer to do the
needed migration steps including possibly necessary manual work.

I will not stop working just because people claim that we should discuss
first and than nothing happens again.  Currently in this regard there
are the following items on my todo list:

  - Enable per binary package citations on tasks pages which could
    be controled by the field "Debian-package" (it has turned out
    that the query became to complex in the current UDD table layout
    and the field is currently ignored)
  - enable more than one citation per package on the tasks pages
    (might be another reason to change the table structure)
  - Enhance BibTeX output
  - Check regularly logfiles of UDD importer and websentinel for
    potential problems
  - Add a "reference of the day" to those Debian Med packages that
    are lacking this information

I'd go much more into details / write better documentation in case
somebody will join this work list.

Please raise your hand if you see any conflict with an other existing
effort or duplication of work to some extend or any other problem in
this approach.

Kind regards


[1] http://wiki.debian.org/DebianScience/ProblemsToWorkOn


Reply to: