Re: Science linguistics and debian-science SVN repository at alioth
Hi Andreas,
Wow, small world! I have been very interested in WordNet for years
and in fact it has been a major (but relatively unpublished) and
recurring interest in my own personal computational linguistics
research. I have made first the popular "poetry generation" programs
using wordnet, and then later wrote programs to search through the
network of synsets for shortest paths to try to glean meaning from
text computationally. Thanks for maintaining this important
package, it has been extremely important in my development and I am
not alone in saying the WordNet package was one of the most inspiring
things to happen in computational linguistics in the 90's.
About the pkg-scicomp version pkg-science debate, I must admit that to
me it is an abuse of English to suggest that pkg-scicomp is somehow
reasonably distinct in meaning from pkg-science. To me, pkg-scicomp
is about Scientific Computation. But anything that you might check
into a "software repository" called "pkg-science" would probably
computer software, and I don't know what software does if not run a
computation. So the naming scheme is bankrupt as it stands and I feel
partially responsible for failing to simplify this complex situation
for our confused users and team members. I have also felt for some
time now that I wished the pkg-science team had a shared repository.
I myself keep dozens of repositories and I do have a lot of open
source projects like CompLearn that I would be happy to move in when
there is a good place to do it. But at this stage I think Debian is
having some flexibility issues and we are sort of stuck unable to
refactor our team organization this year so far. I still think it is
important and overdue and would welcome new shared repository focus.
Best regards,
Rudi
On Tue, May 6, 2008 at 2:58 AM, Andreas Tille <tillea@rki.de> wrote:
> On Tue, 6 May 2008, David Bremner wrote:
>
>
> > Speaking as the maintainer of all packages in pkg-science ( all 3 :-)
> > ), I don't mind any sort of merging plan.
> >
>
> That's good. ;-)
>
>
>
> > But I understood from
> > previous mails from C. Prudhomme that he (and others?) preferred to
> > keep pkg-scicomp a more focussed effort on scientific computation (in
> > the computer science usage of that term).
> >
>
> Couldn't this be apprached by using a subdirectory "computation" ?
> That's what I mean with structure. (BTW, would libblitz++ fit into
> this field or not?) I also have some trouble to draw a really good
> borderline between computation and other fields of science.
>
>
>
> > If that is an accurate
> > statement (Christophe?) then using pkg-science as the umbrella group
> > seems to be consistent with the original intent. It is not completely
> > reasonable to compare the activity in pkg-scicomp and pkg-science
> > since the latter is much younger.
> >
>
> Well, as I said several times: I don't care about the name. I would
> like to see a common repository for people who feel "at home" here on
> this list. If it makes more sense to move those scicomp packages to
> a common repository it is fine for me as well - there is just more work
> to do in this case which might be a show stopper to get it completed.
> I just have a problem to decide what belongs to scicomp and what not
> and thus I would simply broaden the meaning.
>
> An alternative approcah would be to declare scicomp as a separate
> project in comparison to DebiChem and Debian-GIS which is a clearly
> structured group that handles in their own field and properly group
> maintains their packages. This would come close to my overall goal
> to create specific teams for every science which might end in an
> autonomous CDD. My view on Debian-Science is to give those sciences
> a home that are sparse regarding man power to maintain their own
> healthy project. In my view Biology, Chemestry and Geographical
> Information Systems just reached these goal and I would be happy if
> we could "release" more and more children out of the common home to
> grow up independently.
>
> If the latter is the case for SciComp I would like to learn more
> about it, which is the exact scope, are there any subtopics, who
> are the members of the team, etc.
>
>
>
> > If there is to be any sort of reorganization, let me put in a plea for
> > git (hg or bzr are probably also lovely, but I already more or less
> > understand git) as VCS. svn-buildpackage is the suxors when not
> > connected to the internet, and these days my out-of-office internet
> > access is via GPRS (not even 3G). But honestly, I can live with svn
> > if I have to. Tool flamewars are for losers :-).
> >
>
> I was about to write in my initial mail that I would like to stay
> those VCS flamers out of the way if they don't are willing to do
> the work but just want to advertise their pet VCS. If you are the
> one who just did the work and decide for git it is fine for me.
> (Well, I'm perfectly new to git, but have heard good things and
> don't want to block progress - but I would hate a flamewar between
> SVN superior VCSs.)
>
> So if people agree: "Yes we should focus on a common VCS and we
> want to maintain packages as a group in this repository" and your
> are doing the work to implement it it is your choice (Debian is
> a DoOcracy). I just want to hear the point of the SciComp crew
> first whether they feel happy with this move or whether they are
> some kind of "grown up child" out of the Debian Science home
> which works separately on its own (and I wish good luck for this).
>
>
> Kind regards
>
> Andreas.
>
> --
> http://fam-tille.de
>
>
> --
>
> To UNSUBSCRIBE, email to debian-science-request@lists.debian.org
> with a subject of "unsubscribe". Trouble? Contact
> listmaster@lists.debian.org
>
>
--
Which is worse, ignorance or apathy? Who knows? Who cares? --Erich Schubert
Reply to: