Bioconda package for REforge

[jayramr]

Hi Team,

I would need to make this REforge as a bioconda package.

Kindly let me know what are the prerequisities for doing this installation and steps. So i can prepare the staging environment and push to bioconda channel.

Thanks
Jay

[jayramr]

@MichaelHiller Please check

[MichaelHiller]

REforge requires Stubb (with a patch) and newmat.
Stubb (http://www.sinhalab.net/software) and newmat11 (http://www.robertnz.net/download.html)
The installation is described in the README. Important is to patch Stubb, as REforge requires a slighly modified Stubb exectuable.

[jayramr]

Hi,

I'm getting an error

#cd lib/newmat/
#gmake -f nm_gnu.mak
g++ -O2 -Wall -c newmat6.cpp
newmat6.cpp: In member function ‘void CroutMatrix::operator=(const CroutMatrix&)’:
newmat6.cpp:431:13: error: ordered comparison of pointer with integer zero (‘int*’ and ‘int’)
  431 |    if (indx > 0) { delete [] indx; indx = 0; }
      |        ~~~~~^~~
gmake: *** [nm_gnu.mak:10: newmat6.o] Error 1

Kindly let me know the reason for this error.

Thanks
Jay

[MichaelHiller]

This could be a compiler related issue. I haven't seen this error, but I'll ask Bjoern Langer who developed it.

[jayramr]

Thanks @MichaelHiller please update once you get any info

[jayramr]

I tested with compiler gcc/g++ version 10 worked perfectly. gcc/g++ 11 having issues.

Also the documentation for changing directory needed to be corrected. Below is the updated one

patch -p1 < ../../REforge/stubb.patch
cd lib/newmat/
gmake -f nm_gnu.mak
cd ../../
make
export PATH=$PATH:`pwd`/bin
cd ../../REforge/
export PATH=$PATH:`pwd`

Thanks
Jay

[MichaelHiller]

Good point. This depends on where users install and download Stubb (either inside REforge or outside).
I replaced these paths with /path/to/REforge now

[MichaelHiller]

Bjoern mentioned that "maybe the C++ standard changed within the 15 years that this problematic library that Stubb is using is old now and pointer vs integer comparisons used to be allowed but not anymore?"

We are not having gcc 11 on our system, but is there any indication that v11 is not backwards compatible?