Failed to install on Ubuntu
Closed this issue · 8 comments
Hi,
I try to install FLXSA on Ubuntu:
/usr/lib/R/etc/Makeconf:168: recipe for target 'fl__vpa.o' failed
make: *** [fl__vpa.o] Error 1
ERROR: compilation failed for package ‘FLXSA’
- removing ‘/home/george/R/x86_64-pc-linux-gnu-library/3.4/FLXSA’
Warning in install.packages :
installation of package ‘/home/george/Downloads/FLXSA_2.5.tar.gz’ had non-zero exit status
Maybe do you have any idea.
Cheers !
George
Is this from the repository or from github?
I have Ubuntu 16.04 and either option install fine, with just some warnings about char to const conversion.
Can you try
devtools::install_github('flr/FLXSA')
Can you tell me what version of gcc are you using?
gcc -v gives me, at the end of the output
gcc version 5.4.0 20160609 (Ubuntu 5.4.0-6ubuntu1~16.04.5)
OK, I will have to setup a virtual system to test this. I assume some new checks in newer ggc versions are failing on the old codebase of FLXSA. I will get back to you as soon as I am able to devote some time to this.
Some changes to the definition of min and max appear to solve the problem. These changes bring other problems in gvv 4.8.4, so I am committing these changes under the gcc720 branch, for the time being
https://github.com/flr/FLXSA/tree/gcc720
I have not tested the results too much yet, any feedback more that welcome.
You can install using
devtools::install_github('flr/FLXSA', ref='gcc720')
Found a better fix, changing min/max to use std::min/max and the newer malloc. Some type casting along the way, as short gets changed to int if 1 is substracted.
So master branch appears to work now on both versions of gcc. Try with
devtools::install_github('flr/FLXSA')