dev meeting 20220209

Present at the meeting:

• Paul-Elliot Anglès d'Auriac (@panglesd)
• Aya Charaf (@ayc9)
• Sonja Heinze (@pitag-ha)

Current plan

• Sonja finishes creating the ppx-universe (i.e. a workspace with the current ppxlib reverse dependencies) to analyze how big the impact of the 4.14 bump is
• Paul-Elliot improves the way ppxlib itself reports errors
• Paul-Elliot and Sonja work on the documentation/manual
• Aya wraps up the first deriver for the standard derivers project: [@@make]

Meeting notes

Paul-Elliot has wrapped up his work on how ppxlib cleans up the situation when a ppx rewriter raises a located exception and that PR is now merged. So now, merlin features in a file with a raising PPX should work fine. Still, raising interrupts the ppx driver and should therefore be avoided. And ppxlib itself is no exception there! So now, he's going to work on turning the located exceptions raised by ppxlib itself into embedded error extension nodes whenever possible.

Both Paul-Elliot and Sonja have worked on documentation. Paul-Elliot has worked on his section about error reporting and Sonja has written up some information about the Ast_pattern combinators and has written a plan about how to structure the documentation we have better and what to add to it. Both Paul-Elliot and Sonja are going to keep on working on documentation.

Apart from that, Sonja has started creating the ppx-universe for the 4.14 bump and is going to finish that in the upcoming weeks. She's also updated the 4.14-bump PR as well as the branch for 5.00.0+trunk support to make them pass the CI and the opam health check. @kit-ty-kate has made the branch for 5.00.0+trunk support available on opam as ppxlib.0.25.0~5.00preview.

Aya has been refining the make deriver by turning optional record fields into optional function arguments and adding support for [@main] annotations. She has also started to integrate a good error reporting flow, i.e. to embed errors instead of raising. She'll finish that and work on the last two features for make: turning list record fields into optional function arguments and supporting [@default] attributes.

Furthermore, @tatchi has opened a PR at ppx_import that will improve ppx_import's syntax by lifting the AST level at which the ppx_import extension node is attached! Once that work is merged and released (which, of course, will take its time) and some time has passed for ppx_import users to adapt, ppxlib can drop its special ppx_import extension node context.