[Phylobase-devl] Rcpp and OS X compiliation

Mark Holder mtholder at gmail.com
Tue Jan 12 19:06:42 CET 2010


Hi
On Jan 12, 2010, at 11:01 AM, Dirk Eddelbuettel wrote:
>
> On 12 January 2010 at 17:30, François Michonneau wrote:

> |   I did that while I was down with the flu and never got back to  
> it. The
> | package builds and installs but it's not functional. Indeed, the  
> file
> | libncl-2.1.06.so after the install is located in /usr/lib/ncl  
> whereas the
> | software that depends on it looks for it in /usr/lib. Given your  
> experience
> | with packaging, I was wondering if you could indicate me the best  
> way to
> | solve this problem (if you have time/if it can help with phylobase).
>
> --prefix=/usr  maybe?

I'm doubt that will work.  NCL uses a automake's pkglib_ install system.

So it's configure pays attention to --prefix, but the lib goes in $ 
{prefix}/lib/ncl

I don't know how C-extensions to R find dynamic libraries, but I think  
the problem is probably in the getting the appropriate LDFLAGS arg to  
the R-package build so that it knows where to look for libncl.

> |   At least one thing that needs to be changed in the source code  
> of the
> | latest versions of ncl to compile flawlessly with the most recent  
> versions
> | of g++ is to include a few additional headers (you can have a look  
> at the
> | diff.gz file on my ppa).
>
> Yes, I think updating NCL to a sane and current version may well a  
> good plan
> but I can't speak for the phylobase team.

We have definitely fixed some failure-to-include-header bugs that were  
revealed by the latest g++, and have been corrected.

Please let me know if ncl-2.1.08 (which is posted at http://sourceforge.net/projects/ncl) 
  is continuing to give you build problems.
I'd certainly like to correct those errors at the NCL level (even if  
you do decide to continue bundling NCL source code).
The svn repo is at:
	https://ncl.svn.sourceforge.net/svnroot/ncl




all the best,
Mark





More information about the Phylobase-devl mailing list