[Phylobase-devl] [Fwd: phylobase]

[Brian O'Meara]

Thanks for the info, Dirk. I was one of the two people responsible for  
the C++ code in phylobase, and I've consistently had trouble mastering  
build issues (this has come up in other code I've written, too). For  
my part, any patches you could provide to make this more robust would  
be great.

We included Rcpp sources in phylobase because, as you note, during the  
week we were first making phylobase Rcpp wasn't being maintained and,  
more importantly, it was only present on a subset of CRAN mirrors (I  
think we ended up getting it from Italy). Taking out the Rcpp source  
from phylobase and just requiring the Rcpp package makes a lot of  
sense to me, and I will look into doing this after the holidays.

Best,
Brian

On Dec 10, 2009, at 8:37 AM, Ben Bolker wrote:

>
>  The C++ interfacing is a fairly regular irritation for us, it would  
> be
> great to improve it/clean it up -- I'm cc'ing/forwarding this question
> to the rest of the dev team, some of whom may have more informed
> opinions (Brian?)
>
>  cheers
>    Ben
>
>
> -------- Original Message --------
> Subject: phylobase
> Date: Wed, 9 Dec 2009 23:07:13 -0500
> From: Dirk Eddelbuettel <edd at debian.org>
> To: Bolker,Benjamin Michael <bolker at ufl.edu>
>
>
> Hi Ben,
>
> I am writing to you in regards to phylobase for which you are listed  
> as the
> maintainer.
>
> As you may know, I have been working with a very sharp British (now
> grad-)student on automtic builds of .deb packages for all of CRAN.  
> This
> works
> already really well for Debian testing -- see
> http://debian.cran.r-project.org
> We hope to extend that to Ubuntu one day.  Being able to 'apt-get  
> install
> r-cran-ANYTHING' for essentially 2000 values of ANYTHING is pretty  
> neat, esp
> as apt et al really do solve dependencies in a reliable way.
>
> A few packages fail to build, and phylobase is among them.  We build  
> these
> packages (just like normal Debian or Ubuntu packages) in restricted  
> chroot
> and that sometimes triggers erros we do not get in interactive  
> shells. For
> phylobase it is
>
> [...]
> cendants.o phyloXX.o reorderBinary.o reorderRobust.o -LRcppSrc -lRcpp
> -Lncl/ncl/.libs -lncl -L/usr/lib/R/lib -lR
> /usr/bin/ld: cannot find -lncl
>
> and I think has to do with ncl using libtool, and the build running
> configure
> && make but not make install.  So src/ncl/ncl/.libs -- a temporary  
> location
> -- is the only one and that proves too unreliable here.
>
> On the other hand, if one also calls, say, in src/ncl/ncl
>
>    make install prefix=`pwd`/../..
>
> then src/lib/libncl.a is created with about 753kb on an amd64.  We  
> could
> then
> link against this static library which would be easier, and we don;t  
> need
> ld.so or LD_LIBRARY_PATH or libtool magic.
>
> Another thing I noticed is that the packages uses the older  
> RcppSrc.  After
> Dominick stopped working on Rcpp (which he neglected from Nov 2006  
> to Nov
> 2009), I took over maintenance about a year ago.  If you look at the  
> current
> Rcpp package and say its examples/ directory you may get an idea.  In
> essence, I maintained the API but made Rcpp a full-fledged CRAN  
> package.  So
> update.packages() gets you new versions including a new library that  
> is
> ready
> to get linked against -- no more sources of RcppSrc/ in your  
> project, no
> more
> build from configure etc pp.  I'd be happy to walk you through usage  
> and
> builds, and there are other example users.
>
> But fixing the build of phylobase w.r.t. nlc is more important.  If  
> you
> want,
> I can work on a patch, or a tentative 0.5.0.1 that changes the build.
>
> Cheers, Dirk
>
> -- 
> Three out of two people have difficulties with fractions.
>
> -- 
> Ben Bolker
> Associate professor, Biology Dep't, Univ. of Florida
> bolker at ufl.edu / www.zoology.ufl.edu/bolker
> GPG key: www.zoology.ufl.edu/bolker/benbolker-publickey.asc
>
> _______________________________________________
> Phylobase-devl mailing list
> Phylobase-devl at lists.r-forge.r-project.org
> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/phylobase-devl

------------------------------------------------------
Brian O'Meara
http://www.brianomeara.info
Assistant Prof.
Dept. Ecology & Evolutionary Biology
U. of Tennessee, Knoxville