Re: Build Boost Back Better

From: Matt Borland (matt_at_[hidden])
Date: 2022-04-06 01:28:46

I would like to throw out some thoughts as one of the newer repeat contributors.

1) The barrier to entry is really high. John Maddock still points out macros or small pieces of legacy wizardry to me, and it's been over 2 years since my first contribution to Boost.Math. There is also the issue of unmaintained libraries. I submitted some trivial changes to Boost.Random that would fix breakage in Math, and it took month's until John overrode the supposed maintainer and merged them in. I think that something to the affect of a Boost version of Kernel Newbies would pay dividends in onboarding new contributors. John, Chris Kormanyos, and Nick Thompson have all been hugely helpful in getting me up to speed but I know it took a lot of time and effort on their end.

2) As others have mentioned I think there is a lot of good conversation that exists in other areas that the mailing list. The mailing list isn't particularly high volume (I have this, the module discussion, and Math/Multiprecision internal as the only 3 active) so I don't know if moving to a forum helps. There also generally seems to be no clear consensus when things get posted to the Mailing List that impact everyone. The one that comes to mind is the discussion of moving on from C++03. I am 29 which means C++11 was finalized when I was a freshman in college. You are not going to attract new people by making them use ancient standards. I appreciate that Math and Multiprecision made the internal decision to move a minimum of C++11 with additional features using 17 (e.g. parallelization).

3) Vinnie posed the questions how do we get more people involved as review managers? I will ask in return how does one get involved as a review manager? I may not be an expert on his proposed URL library, but I certainly have experience in project management. The webpage describing the process seemed a bit nebulous to me, and I am likely not alone in thinking this.

4) Another thought on deepening involvement for people like me is there a list of unmaintained libraries that need someone? What is the process for becoming a maintainer of those or in general? I am pretty sure the only reason John gave me write access to Math was so that I would stop pestering him. What help do other libraries need and what is their glide path? I have spent the better part of the last year and a half working on standalone modes for Math and Multiprecision. The standalone math is now used in the MSVC STL and Multiprecision was done by request. We have all seen the Reddit posts about people not using Boost because it is a monolith so we know there is a large demand signal. If there are other libraries that want to pursue this I am certainly willing to help. I also think an information campaign on moving beyond the monolith would be worthwhile.

Again just my 2 cents. I do not intend to undermine the work that has been going on since I was a child.

Matt