using cmake for building RMINC

[vfonov]

Hello,

I converted configuration part to use cmake, take a look in https://github.com/vfonov/RMINC/tree/cmake take a look, if it's usefull for you.

[cfhammill]

Thanks Vlad, it certainly looks much cleaner than the old build system. I'm hesitant to incorporate it into master in the foreseeable future, https://github.com/stnava/ITKR/issues/1 and https://github.com/stnava/cmaker/issues make it seem as though using cmake is instant disqualification from CRAN at this point.

[vfonov]

https://cran.r-project.org/web/packages/qtbase/index.html - is cmake based, for example.

[cfhammill]

hmm, so it may be possible. Still don't really want to tempt fate with the CRAN maintainers, but maybe after the first CRAN release I'll switch the build system and see if they care.

[vfonov]

I am making more changes: https://github.com/vfonov/RMINC/tree/minc2-simple , using my new library for MINC2 IO ( https://github.com/vfonov/minc2-simple ). So, running benchmark : ( https://github.com/vfonov/RMINC/blob/minc2-simple/etc/benchmark2.R )

> benches
      test replications relative elapsed
1   mincLm          100    1.000  18.073
3 fLmSEXP2          100    1.158  20.936
2  fLmSEXP          100    1.232  22.258
4   lm.fit          100    1.403  25.349

fLmSEXP2 - is using https://github.com/vfonov/RMINC/blob/minc2-simple/src/minc_apply2.cpp#L46

[cfhammill]

Nice, the code for mincApplyRCPP2 looks cleaner too. I now understand the push to switch to cmake.