[R-pkg-devel] R CMD build with configure-args

[Russell Almond]

Thanks for the suggestion.  I do have a configure script (that is what 
is reading the --with-netica arg) but I didn't think about searching the 
standard install locations.  I can probably use that to work around this 
problem.

However, that doesn't completely solve the problem.  If I want to force 
the configure script to use a non-standard location (say to use an 
experimental version of the library), I still need to be able to pass 
configure args to INSTALL and check.  I admit the need for doing this 
with build is a little bit less clear, although it is also unclear why 
build (which is making a source tarball) needs to compile the C code.

	--Russell


On 07/30/2015 05:11 PM, Dirk Eddelbuettel wrote:
>
> Besides what Duncan said, relying on user to supply arguments is pretty bad
> as it more or less guarantees _any_ automated test will not succeed (for lack
> of involvement of the sage user).
>
> Writing configure scripts feels like yet another painful step, but it really
> is not that hard if you know a little shell scripting already.  You can even
> do simple things and just loop over a fixed (but hopefully long) list of know
> locations -- which is what RPostgreSQL does:
>
>    https://github.com/cran/RPostgreSQL/blob/master/configure.in#L38-L69
>
> Not my proudest moment, but it worked for a few (ten-)thousands of
> installations and builds ...
>
> Dirk
>

-- 
Russell Almond
Associate Professor, Measurement & Statistics
Educational Psychology and Learning Systems
1114 W. Call St.
Florida State University
Tallahassee, FL 32306
(850) 644-5203
ralmond at fsu.edu
http://ralmond.net/