include header files for large libs?

[Jeffrey Sarnoff]

As an example, I have a package that uses a C library which depends on the presence of libgmp and libmpfr, both of which are large and come with Julia.  On my Linux,  libgmp.so and libmpfr.so are found in /usr/local/lib/julia/. If libgmp.h and libmpfr.h were found in /usr/local/include/julia, then I would not need to install duplicates or go through machinations to determine which version of each is in use and find the appropriate headers.

[Andreas Lobinger]

Hello colleague,

On Wednesday, October 5, 2016 at 9:37:29 PM UTC+2, Jeffrey Sarnoff wrote:

As an example, I have a package that uses a C library which depends on the presence of libgmp and libmpfr, both of which are large and come with Julia.  On my Linux,  libgmp.so and libmpfr.so are found in /usr/local/lib/julia/. If libgmp.h and libmpfr.h were found in /usr/local/include/julia, then I would not need to install duplicates or go through machinations to determine which version of each is in use and find the appropriate headers.

there were discussions before, if and how many external tools julia should ship with on certain platforms (like git on windows...) and one clear message was "Julia is not a distribution mechanism for X". There might be reasons, why your julia installation on your linux box includes certain .so-s but actually local lib builds in julia should be the special case (I'm building LLVM, which takes quite some time). If you need a certain library as infrastructure, you need to take care it exists.

[Jeffrey Sarnoff]

I understand your point.  My post was motivated by the fact that I can (indeed should) use whatever versions of libgmp and libmpfr go with Julia.  Doing that makes potential interoperability issues vanish.  However, if there is a policy in place that prefers to avoid that -- nevermind?

[Scott Jones]

I think that would be rather problematical, as those are source files, which might cause licensing issues for Julia, as the GMP and MPFR libraries are under the LGPL.

[Jeffrey Sarnoff]

good point.  I have adopted a different approach.  never mind :)