[pkg-bioc] Request for description of setup mirroring etc

[Dirk Eddelbuettel]

On 19 May 2005 at 09:25, Egon Willighagen wrote:
| On Wednesday 18 May 2005 09:51 am, Steffen Moeller wrote:
| > Egon Willighagen wrote:
| > >>For Bioconductor we have not addressed the issue of these data packages
| > >>yet. We certainly need them only once.
| > >
| > >What do you mean with 'only once'?
| >
| > I thought that since the packages are updated fairly regularly, an
| > update of the automated build procedure can easily wait until the next
| > release of an upstream package. To reserve the local disk space for the
| > CRAN mirror would then not required.
| 
| IC. Indeed.

I had planned to work on that part "one day". I think a summary of which CRAN
packages went into the archive on a given day, what their versions are etc
would be useful. This could either trigger an email, and may equivally well
trigger a built of new packages if tied into our to-be-completed system.

Easy enough to read the status from "database" (of whatever form) remotely,
not so sure about writing back.  Any suggestions?  Csv file and scp push?

| > How irritated would you be, Egon, if we merged cran2deb and bioc2deb
| > again? 
| 
| Not at all. I haven't written the scripts, only hacked in the cran2deb.pl...
| And I don't even know the exact details why there are split...

Oh, they should be merged. But that is one goal, making either one smarter is
another...

| > Well, I'll start with the CRAN and then see how BioC works and 
| > come back to you.
| 
| Ok. There's a glitch in cran2deb.pl... Recently, I've added some logging of 
| the changelog, which works, but there needs to be added some detection if a 
| new deb package needs to be build based on version numbering of cran2deb.pl 
| and CRAN2DEB.pm...

See above -- I think it needs static memory.
 
| > However, somehow I feel this separation to be very artificial.

It was a practical consideration at the time. CRAN provided its information
in ways that are (were?) different from BioC. The reposTool package helps as
it unifies the viewing, but then CRAN is treated via its mirror on BioC.
 
| How important it is to merge them?

Somewhat, but I wouldn't call it critically important (aka "drop all tools
and do it first before we continue").

My $0.02.

Cheers, Dirk

-- 
Statistics: The (futile) attempt to offer certainty about uncertainty.
         -- Roger Koenker, 'Dictionary of Received Ideas of Statistics'