Re: Packaging scientific datasets for Debian
On Mon, May 10, 2010 at 11:05:33PM +0900, Charles Plessy wrote:
> Le Mon, May 10, 2010 at 06:05:04AM +0200, Joerg Jaspert a écrit :
> > On 12109 March 1977, Steve M. Robbins wrote:
> > >> http://ftp-master.debian.org/wiki/projects/data/>
> > >> Although it makes the impression that everything is already done, I
> > >> don't know if that is actually true. Does anyone know about the current
> > >> state of this effort?
> > > I don't know. But given that data.debian.org doesn't resolve, I'd
> > > guess that nothing is set up. The wiki page you reference suggests it
> > > is an ftpmaster team consensus position, so I cc them now. Maybe
> > > someone there can chime in.
> >
> > We are waiting for hardware with enouggh diskspace. This is aboout there
> > (ftpmaster.d.o replacement).
>
> Hello Jörg,
>
> I read in http://ftp-master.debian.org/wiki/projects/data/ that the data
> archive will require full source uploads. But if the source package is not a
> simple downloader, this will duplicate the data and double the size of the
> upload and the archive. Would it be possible to accept binary-only uploads? I
> ask the question in particular because with one of our tools, getData, I am
> considering to produce binary pakcages from scratch (or with a helper tool like
> equivs, for instance). That way, the data package can originate from a signed
> official package distributed in the main archive, but does not need a Debian
> source package.
For unoffical packages I have used a similar approach: an almost empty
source package that build-depends on the binary packages (with exact
version) that it is supposed to build. This was once suggested by
Anthony Towns:
http://lists.debian.org/debian-devel/2007/06/msg00298.html
However, FTP master previously expressed dislike of this approach. So
I'd rather ask whether source-only uploads would be possible?
Packages in question are in the GB range, so the difference is 1 vs 2 GB
to be uploaded. Since data.d.o is maintained by the project I'm now less
concerned about archive size -- although I still consider it wasteful to
duplicate identical data in two different formats (tar.gz, deb).
Michael
--
GPG key: 1024D/3144BE0F Michael Hanke
http://mih.voxindeserto.de
Reply to: