Consider using swig-fortran to access C++ STL

[rweed]

I'll just throw this out as an alternative to building stdlib from scratch. Seth Johnson at ORNL has a project that can generate SWIG bindings to the C++ STL for Fortran. (see https://github.com/swig-fortran). I ran across his project a few months back and thought it interesting but didn't do a deep dive into what issues where involved in using it for things like lists etc. More info in the following two PDFS
Johnson_Automated_Fortran-C++_Bindings_ArXiv.pdf
siam-cse-johnsonsr.pdf

[certik]

STL can help with the containers and algorithms part of stdlib. I don't think it would help with any of the current functionality in our current master, would it? Or with the mathematics part. See the README for the intended scope of stdlib. Also if the C++ dependency could be only optional would be also helpful to lots of future users of stdlib.

…

On Sun, Jan 5, 2020, at 9:30 AM, rweed wrote: I'll just throw this out as an alternative to building stdlib from scratch. Seth Johnson at ORNL has a project that can generate SWIG bindings to the C++ STL for Fortran. (see https://github.com/swig-fortran). I ran across his project a few months back and thought it interesting but didn't do a deep dive into what issues where involved in using it for things like lists etc. More info in the following two PDFS Johnson_Automated_Fortran-C++_Bindings_ArXiv.pdf <https://github.com/fortran-lang/stdlib/files/4023624/Johnson_Automated_Fortran-C%2B%2B_Bindings_ArXiv.pdf> siam-cse-johnsonsr.pdf <https://github.com/fortran-lang/stdlib/files/4023626/siam-cse-johnsonsr.pdf> — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#84?email_source=notifications&email_token=AAAFAWE7OYLIVSQHT6H7JVTQ4IDKVA5CNFSM4KC4AFU2YY3PNVWWK3TUL52HS4DFUVEXG43VMWVGG33NNVSW45C7NFSM4IECXI4Q>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAAFAWF5FXFROCEYK6KOICDQ4IDKVANCNFSM4KC4AFUQ>.

[rweed]

Actually, I personnaly prefer a "pure" Fortran implementation if possible. I just put this forward as something that could be an option to provide some of the containers etc. needed for ADTs etc. if there is no consensus as to the best way to do it in Fortran. I agree it should only be an option (and only if its impracticle or time-consuming to do it in Fortran).