[pkg-bioc] how to proceed?

[Dirk Eddelbuettel]

On 19 February 2005 at 20:28, Egon Willighagen wrote:
| I'm not a Debian expert but /var/spool sounds the right place... apt stores 
| the debs there too...

That's a change we need to make, yes.

| > I also have a bioc2deb.pl partial rewrite that uses more R for information
| > retrieval (given BioC's excellent and very feature rich repostool package).
| > I never cleanly/fully ported that back to cran2deb.pl, though.
| 
| I've briefly looked at the code... i've seen this information in the .pl file:
| - some base and recommended package names

Yes, because they get special treatment
i)   we don't need to (re-)build them
ii)  Depends of R packages are treated differently. Base packages are given,
     and recommended ones are treated differently too (can be skipped at
     build, yet need an explicit depends)

| - some package dependencies

iii) Bundles need a mapping too: a Depends on MASS needs to be mapped to
r-cran-vr. That may need to be done similarly for the dse and gremisc bundles.

| - a number of packages that cannot be build (missing dependencies)

Yes, or fail to build.

| Is there more things to configure?

These are 

| Why are base and recommended named separately? Just to know to build those 
| first?

See above -- you need explicit depends on recommended packages, but can
assume base to be installed with R.

| BTW, how did you make a mirror of just the packages instead of the whole of 
| CRAN?

I don't understand the question. I downloaded all source packages from CRAN,
plus one or two of the meta files that have all the information. Those are
parsed mostly with Perl tools included in R.

BioC has its own mechanism with reposTools.

It would be nice if you (or someone else) could try to run the script and see
what happens. You will probably need to update things like build
dependencies.

Dirk

-- 
Better to have an approximate answer to the right question than a precise 
answer to the wrong question.  --  John Tukey as quoted by John Chambers