Make good faith effort to detecting system installed normaliz before falling back to compile our own

[fingolfin]

Users who have a recent Normaliz installed should not be forced to tell NormalizInterface explicitly so; rather, we should just detect and use it.

[alfsan]

Maybe this can do the work. In case the user gives an argument with the directory where Normaliz is installed, it tests in typical locations.

LOCATIONS="/opt /usr/local /usr"

nmzlocation()
{
  if [ $# -eq 1 ]; then
    LOCATIONS="$1 $LOCATIONS"
  fi
  for l in $LOCATIONS; do
    if [ -f $l/include/libnormaliz/version.h ]; then 
      NORMALIZ_DIR=$l
      break
    fi
  done
}

and get the version:

NORMALIZ_VERSION=`sed -rne 's/(.define [A-Z]+_VERSION[[:space:]]*) ([0-9])/\2/ p' ${NORMALIZ_DIR}/include/libnormaliz/version.h`

[fingolfin]

Maybe. Somebody needs to invest the work and implement it, then test it in various combinations. Pull requests are welcome :-)

[alfsan]

I created a fork from gap-packages/NormalizInterface. In the fork I have a branch called "detect-normaliz" with the changes to try to detect Normaliz.

Is it Ok if I create a pull request from the branch "detect-normaliz" in the fork, to the master branch of gap-packages/NormalizInterface?

[fingolfin]

@alfsan of course you can always create pull requests, that's one of the main reasons for having this code on GitHub in the first place :)

[alfsan]

@fingolfin I will wait to the resolution of PR #79 to do a new PR incorporating the changes in "detect-normaliz", mainly because I don't know if I can finish breaking something

[alfsan]

@fingolfin previously to PR #79 I created a (local/fork) branch with the intention to address the present issue. Is it OK if I present a pull request for this issue, based on the NormalizInterface state before PR #79, or should I proceed in a different way?