[Rd] Re: HDF5 as a module

[Kurt Hornik]

>>>>> Marcus G Daniels writes:

> Included below are:

>   1) Diffs to to remove HDF5 from R

>   2) Diffs to pass thru --with-* options to a module's configure
>      script during "R INSTALL".  (To reiterate a plea for this
>      capability...)  This applies only to Unix, the Windows (mingw32 in
>      my case) build approach of passing in make variables like
>      *-CFLAGS and *-LDFLAGS works fine.  The only hitch with the
>      Windows install is that I had to drop the `bundle' wrapper, which
>      was useful for Unix in order to give the package a recognizable
>      name with version number (Rhdf5-0.0), but the installed package
>      with R a minimal name (hdf5).

>   3) The module itself, as an attachment.

I made the changes to the R sources, and removed the obsolete files.

Re INSTALL, I had already committed the new `--configure-opts=' option,
which I still think is preferable to having `--with-*' explicitly.

Re the module, I have several comments.

* It would be nice to merge this into Doug Bates' foreign package.

* I think this should be one package, not a bundle.  Also, the version
info in DESCRIPTION does not match the actual number used.

* Makevars.in should not use LDFLAGS.  Instead, the -L flags should be
in PKG_LIBS (as explained in R-exts.texi).  Also, linking against the
libraries should only happen if these were found.  In general, I am not
sure how the package will install in case the libs are not there.

* The include statements in hdf5.c should conform to the new header file
structure.

-k
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-devel mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-devel-request@stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._