Re: R packages

[brook%biology.nmsu.edu@localhost (Brook Milligan)]

Greg,

Thanks for your comments.

Greg Troxel writes:
 > It would be nice if in the glorious future there was a version of this
 > for other languages (as Benny mentioned CPAN), and even nicer if there
 > was some common infrastructure for the
 > perl/tex/python/R/ruby/guile... variants.

Yes, but one step at a time.

 > I generally think it's a feature that the script is not interactive, but
 > placing of new packages seems tricky.  Still, either way is a huge
 > improvement over the way things are.

Any suggestions on how to specify package-specific categories in a
non-interactive and not-too-cumbersome way are welcome.  Given current
practice of everything being in 'math', though, I'm not sure how much
to worry about this.

 > It sounds like this can regularize Makefiles for R packages that are
 > different from how they'd be generated, and with manual review that
 > could lead to cleaner pkgsrc entries; if I followed that right it sounds
 > sensible.

If new packages are created with the tool, then they will have a
consistent structure.  I have avoided the challenge of interpreting
too much of the Makefile structure when updating packages, though, so
regularizing them will only happen with time.  Doing otherwise risks
errors when handling all the odd cases of package-specific
idiosyncracies concerning special variables, etc.  Luckily, the R
packages as a group are fairly consistent already.

 > I don't see any reason not to add pkgtools/R2pkg with the program.

Good.  Given that I still have to write an updated man page and clean
up the code, it will be a bit.

Thanks again for the input.

Cheers,
Brook