Catalyzing gfortran development

Should/could the Fortran community incentivize the gfortran development and bug fixes? And if yes, how and where should we begin? I am particularly interested in accelerating the full implementation of Parameterized Derived Types in gfortran. In my opinion, the full development of this particular feature has dragged on for too long. Other Fortran compilers provide full implementations of PDTs. Still, I am hesitant to use any language feature that gfortran cannot compile (for several good reasons, portability is one of them). Currently, I do not see any benefits in a fully functional implementation of the len type parameter of PDTs, if it is not performant. However, the kind type parameter is a compile-time constant and critical for generic programming involving derived types. This is currently implemented in gfortran, but the implementation is plagued with several major bugs that render PDTs useless.
Unfortunately, I cannot directly get involved in the development of gfortran (for the same reasons I cannot repair my car or can not do backyard gardening). Still, if we can figure out a way to catalyze the gfortran development, I am happy to contribute financially to such efforts personally.

8 Likes

Coopting a different slogan, “Free software isn’t free”.

3 Likes

Brings back memories of a co-worker. She used to have a laugh-box on her desk that she inserted little quotes into a panel you glued to it. The “…someone will fix it” was on there. If you picked it up it played back hysterical laughter. She also had a really nice sign made in her office of the old saying “I heard you want it cheap, fast, and bug-free – pick two”. Might not quite have the quote right, but it is close.

Similar discussion came up to a project I participate when I found out service like https://bountysource.com/ or https://issuehunt.io/ (disclaimer: I have no relation to them) where issues can be funded with financial reward.The bounty can be posted by anyone who wants to see the issue resolved. If @shahmoradi would like to contribute financially this is a legit option.

1 Like

So it is a bounty for a specific feature versus a general fund with a goal like “make gfortran f2018-compliant” or “support gfortran”? is it done primarily based on bugzilla entries? Surprisingly, I have not found much information on the general topic of funding gfortran so far.

Yes the bounty is supposed to for specific issues, and [gfortran’s PDT has a lot] (82173 – (PDT) [meta-bug] Parameterized derived type errors). I assume it can be done based on any issue entry (bugzilla, github, gitlan, etc), tho not 100% sure.

I just want to note that one cannot just jump into a compiler project and expect to be productive quickly. It can take months, with handholding by experienced project members, to begin to understand how various parts work. Heck, I had years as a developer on VAX FORTRAN and it took me quite a while to be able to contribute meaningfully to DEC Fortran, which was a completely different codebase. Even then, my knowledge was confined to specific parts of the compiler.

The older the project gets, the harder it is to get up to speed and understand all the moving parts.

6 Likes