[pkg-bioc] Hello!

Andrew Dai ad364 at hermes.cam.ac.uk
Fri Jun 9 07:55:58 UTC 2006


Hi all,

My exams finally finished yesterday so I should be able to start work on
this soonish. Personally I'd prefer to code at least at the moment in Perl,
partly for the reason that R's supporting scripts are all in Perl themselves
and partly because of all the pre-existing code. I'm happy with splitting
the small tasks using Python.


Andrew

-----Original Message-----
From: Johannes Ranke [mailto:jranke at uni-bremen.de] 
Sent: 09 June 2006 05:14
To: Dirk Eddelbuettel
Cc: pkg-bioc-devel at lists.alioth.debian.org; Andrew Dai
Subject: Re: [pkg-bioc] Hello!

> On 8 June 2006 at 17:19, Johannes Ranke wrote:
> | Thanks! I think the alioth pages really should be improved a little 
> | bit (see my original mail).
> 
> Go right ahead :)

I can't, at least as far as
	
	http://alioth.debian.org/projects/pkg-bioc/

is concerned. In the summary text, CRAN is not even mentioned!
 
> | You mean the Wiki page AliothPkgBioc? Yes, this could be clarified.
> 
> Idem. I think Rafael may have made the wiki-page our home page. Not a bad
idea.

Yes, that's nice!
 
> | Some related questions: Is somebody really using all these packages 
> | that can be built? Do you use cran2deb.pl for your Debian packages? 
> | It sounds like you use an improved version on your machine for this.
> 
> I used 'my' first version of Albrecht's / Detlef's script for an 
> earlier Quantian build. Now for Quantian I just install into /usr/local.
> 
> So I don't think anybody uses the packages yet -- but it's a vicious 
> circle because we also don't run the script regularly yet.

Maybe we should scale down the scope of the CRAN script to packages that
actually build and are useful in Debian and that we care for. There is a lot
of CRAN packages now that are probably only interesting for a very small
number of people, as far as I can see.
 
> | So there would be two possibilities: 
> | - Upload everything to Debian (which I would also be hesitant to do,
> |   although this list could try to deal with the BTS).
> | - Maintain a separate repository, maybe even on CRAN and BioC servers.
> 
> We don't yet. We also think 'all to Debian' is to massive. Maybe one 
> day after we have a bit of a track record and more experience outside 
> of it. I am sure we will find repo space somewhere -- as you said,
possibly even on CRAN.

I can use my own repo for testing things in the meantime, but it's not split
into stable, testing and unstable (yet).
 
> | <off-topic>
> | I would prefer python, since I would like to learn it, as opposed to 
> | perl, which seems outdated to me (although I have a strong 
> | inclination to Debian, which is intimately tangled with perl, and 
> | think it is a modern distribution. We had KDE 3.5.3 packages in sid 
> | even before it was released!).
> | </off-topic>
> 
> Even if you rewrote it all we had the problem that the current crew is 
> using Perl -- there is some subtle self-selection going on.  But 
> longer term maybe we can break the script into several smaller tasks 
> (updating the source mirror, updating the metadata, computing the 
> mapping to Debian packages, publishing results etc pp) some of which may
well be Python-based.

I think splitting things up would be a good idea - maybe with a simple
makefile to coordinate the smaller tasks.
 
> But for now Perl it is I'm afraid.  [ I have also meant to learn more 
> Python, now for decade and counting ... Perl just "works" for me. ]

What about Andrew - what do you prefer?

Best,

Johannes




More information about the pkg-bioc-devel mailing list