ppx_sexp_conv v0.11.1

[ghost]

Remove the dependency from ppx_sexp_conv.runtime-lib to base

[camelus]

✅ All lint checks passed 7142f8f

• These packages passed lint tests: msgpck.1.0, ocaml-topexpect.0.1, ocaml-topexpect.0.2, ocaml-topexpect.0.3, ppx_sexp_conv.v0.11.1, qcow.0.6.0, qcow.0.7.0, qcow.0.8.1, qcow.0.9.0, qcow.0.9.4, qcow.0.9.5, yaml.0.1.0

---

✅ Installability check (8793 → 8794)

• new installable packages (1): ppx_sexp_conv.v0.11.1

[hannesm]

thanks. are the revdeps failures expected? I suggest we merge this, and afterwards a PR similar to #11898 which (a) conflicts with ppx_sexp_conv {= "v0.11.0"} and (b) turns the build dependency ppx_sexp_conv {build} into a runtime dependency ppx_sexp_conv. if I follow mirleft/ocaml-nocrypto#144 along, there is as well the requirement to expose this runtime dependency via META!? since when does ppx_sexp_conv provide a runtime-lib (i.e. do we need as well a lower bound of ppx_sexp_conv in the respective packages?

[ghost]

Looking at the errors, some of them should already occur with ppx_sexp_conv v0.11.0, I'll add some constraints. Some are just missing a dependency on sexplib.

Basically, the runtime dependency of ppx_sexp_conv used to be sexplib, but there was a change between v0.10 and v0.11 to add a ppx_sexp_conv.runtime-lib library and make it the runtime dependency of ppx_sexp_conv. This runtime library simply re-exports some functions under a Ppx_sexp_conv_lib toplevel module and the generated code refers to it. This was part of a change that affected all our ppx rewriters and was motivated by:

• keeping the generated code as small as possible and move more code to runtime libraries, since it's much easier to review and maintain
• hygiene: we had some name clash problems and by systematically having a runtime library we have a better guarantee on what the identifiers produced by ppx rewriters refer too

[ghost]

For shared-block-ring and vchan-unix the issue seems to be unrelated to this PR so I didn't touch them.

@avsm and @let-def, after this PR, ocaml-topexpect and yaml needs to be updated to add an explicit dependency on sexplib. For ocaml-toplexpect, the sexplib dependency was already in the jbuild file so I just added it to the opam file, there should be no need for another release, but for yaml, I think we'll need a new release.

[ghost]

Not sure what's going on with the CI here, I constrained things more so there should be less failures... The new errors look unrelated to this PR

[ghost]

Ok, it looks better now. I tested manually and ocaml-topexpect 0.1, 0.2 and 0.3 don't build with OCaml >= 4.05, so I added more constraints. The last two failures - qcow 0.6 and 0.7 - are unrelated.

I think this PR is good to merge.

[ghost]

Actually I added the missing dependency on io-page-unix for qcow 0.6 and 0.7 and the result is a bit better now. The last failures are really unrelated to this PR

[hcarty]

Agreed that this seems ready to merge.

@hannesm Do you want to take a look since you were reviewing this?

[trefis]

Ping, it would be nice to unsplit the world ASAP.

(x-ref mirage/ocaml-uri#120 )

[trefis]

Forget about the "unsplit the world part", apparently my opam has just gone crazy on its own.
Still, it would be nice to get this released.

[hannesm]

sorry for the late reply. This is good to go. to unsplit the universe we'll still need to carefully adapt packages to the new ppx_sexp_conv world (thanks @diml for explaining the design considerations).