dev meeting 20210211

Present at the meeting:

• Jérémie Diminio (@jeremiedimino)
• Sonja Heinze (@pitag-ha)
• Nathan Rebours (@NathanReb)

Current Plan

• Nathan creates a post on discuss to announce the new release and report on the upgrade process
• Sonja keeps working on Astlib

Meeting notes

Nathan worked on mdx to make it compatible with the latest ppx eco-system: it was still relying on omp1 which made it non co-installable with ppxlib. A PR had been opened a while ago to port it to ppxlib but Guillaume its original author ran into some issues and requested help to finish it. After a bit of investigation it turned out that mdx could easily work with its own thin compat layer, using OMP or ppxlib there was not the proper solution. Carl confirmed this new OMP-free MDX worked within the JS monorepo and so Nathan released this new mdx.1.8.1. Nathan haven't had the time to announce the latest ppxlib release on discuss and will take care of it by the end of the week.

Sonja got back to work on Astlib, mostly rebasing her branch on top of master since that's been a while since she last got to work on it. Doing so she realized that it wasn't working when the compiler and ppxlib AST were different so she's working on fixing it.

When discussing the fix for MDX we talked about one of the points the original ppxlib port PR was blocked on. The Location.Error exception defined in ppxlib is mostly unused, e.g. ppxlib's Location.raise_errorf doesn't raise it but the compiler-libs variant instead. When porting to ppxlib it is sometime the case that you try to catch location errors but fail to do so because you have Ppxlib opened and let the compiler-libs one slip through as a result. We need to adress this in the future but it will most likely prove simpler as Astlib will have to define a stable location exception, the one raised by the parser. We'll keep the status quo for now as there are known work arounds. Once Astlib is out we'll have a proper solution.