[pkg-bioc] cran2deb.pl error on gmodels

[Johannes Ranke]

* Dirk Eddelbuettel <edd at debian.org> [070829 14:10]:
> 
> On 29 August 2007 at 13:09, Egon Willighagen wrote:
> | On 8/29/07, Egon Willighagen <egon.willighagen at gmail.com> wrote:
> | > I'm also thinking of separating these bits of information from the
> | > source... will get back on this.
> | 
> | OK, submitted a patch for this (r308). %cannotBuild is now read from a
> | separate file: tools/cannotBuild.list
> | 
> | Will do the same for other bits of information.
> 
> Before we go completely overboard here in moving information around without
> adding anything fundamentally new, it may be worth figuring exactly how we
> want to do that.  
> 
> I still think a shared database is best.  Given that we all have _write_
> access on alioth, we could try to cook something up with simple locks that we
> set whenever we change values, and otherwise _read_ by rsync/scp/... and
> re-reading the db.  Which could be flatfiles, or sqlite.

I don't see the advantage over plain text files being read directly by
cran2deb.pl. Nowadays searching in a text file is so easy and fast (e.g.
using the vim editor :). A real advantage in my eyes would be to have a
dependency tree, e.g. in an R object created by the package that you
mentioned a while ago.

> And just to be a stickler: the bug is that the code doesn't figure that
> gmodels/gdata are not needed. Re-declaring that in 'cannotBuild' is logically
> wrong ("cannot" is false, "not required to build" is true). We need to find
> the bug.  Unfortunately, the code is murky...

I second this. The packages that are in Debian should be separated from
the ones that can't be built. The contents from the former could then be 
extracted from the debian archive by apt-cache searching for r-cran-*,
r-bioc-* and r-omegahat-*.

> Ok, enough complaining.  Let's celebrate that we have a minor amount of
> momentum. 
> 
> As for the sponsoring question: Sorry, I can't, too many packages already.
> 
> But good news is that that Simon tells me that rJava, JGR, ... all build with
> the free java tools and could go into main.  Graphical parts still work
> better with the non-free Java, but they do not need the CRAN package to be
> rebuilt: just (re-)installing the commercial Java and re-running 'R CMD
> javareconf' is all it takes.  So this may help for jmol...

Good news indeed. BTW, rkwards has improved a lot too, lately.

Hannes

> 
> Dirk
> 
> -- 
> Three out of two people have difficulties with fractions.

-- 
Dr. Johannes Ranke                 jranke at uni-bremen.de     Key ID: F649AF90
UFT Bremen, Leobenerstr. 1         +49 421 218 63373
D-28359 Bremen                     http://www.uft.uni-bremen.de/chemie/ranke