6 packages from bensmrs/gendarme at 0.3.0

[bensmrs]

This pull-request concerns:

• gendarme.0.3.0: Libraries to marshal OCaml data structures
• gendarme-json.0.3.0: Libraries to marshal OCaml data structures (JSON)
• gendarme-toml.0.3.0: Libraries to marshal OCaml data structures (TOML)
• gendarme-yaml.0.3.0: Libraries to marshal OCaml data structures (YAML)
• ppx_marshal.0.3.0: Preprocessor extension to marshal OCaml types
• ppx_marshal_ext.0.3.0: Preprocessor extension to simplify writing Gendarme encoders

---

• Homepage: https://github.com/bensmrs/gendarme
• Source repo: git+https://github.com/bensmrs/gendarme.git
• Bug tracker: https://github.com/bensmrs/gendarme/issues

---

🐫 Pull-request generated by opam-publish v2.5.1

[bensmrs]

What would you recommend me to do in this case? Drop 4.13 support? Reimplement Seq.equal myself? Is there a mechanism in Dune/Opam to provide custom implementations for OCaml versions where such implementations are missing?

Thanks!

[raphael-proust]

Drop 4.13 support? Reimplement Seq.equal myself? Is there a mechanism in Dune/Opam to provide custom implementations for OCaml versions where such implementations are missing?

There are some compatibility packages to bridge that gap. You can give https://github.com/c-cube/seq/tree/master a try maybe? I'm not sure it'd work for your usecase, it might be for the 4.07 gap instead. Maybe https://github.com/ocamllibs/stdcompat?

Otherwise, there is a mechanism in dune to include modules conditionally (see enable_if) and a mechanism to select different implementations (see https://dune.readthedocs.io/en/stable/reference/library-dependencies.html#alternative-dependencies).

Dropping support for 4.13 is also a valid choice. You can maybe query opam to check which of your revdeps require ocaml<4.14 to help you decide if that's the correct thing to do.

[raphael-proust]

Looking at the actual error, it seems to only affect tests. Maybe you can deactivate those tests for <4.14?

[bensmrs]

Hi!

Thanks for those very nice pointers.

Looking at the actual error, it seems to only affect tests. Maybe you can deactivate those tests for <4.14?

Oh you’re right! I got mixed in my head because I force-pushed a fix removing the need for Map.S.of/to_list in this PR, so the question can be seen as a “good to know for my future self”.
I’ll have a go with Stdcompat for the tests then.

I asked that in the past, but it got lost in plenty of other questions of mine. In such cases where I’m using opam publish, is closing the PR and opam publishing again the way to go or is there a more concise way supported through Opam to force-push on a PR?

[bensmrs]

Well turns out I was too quick closing my PRs and opam publish does forcepushes just fine. I’ll be damned!