Added cloudi.2.0.7

[okeuday]

No description provided.

[okeuday]

Only macos-homebrew-ocaml-5.1-arm64 (experimental) and macos-homebrew-ocaml-4.14-arm64 (experimental) failed due to not having GMPlib for zarith to use.

[mseri]

Strange, gmp should be there. I think homebrew uses a different path on arm, maybe the build script is looking in the wrong place

[okeuday]

@mseri The failure occurs in the zarith-1.13 release that is included in the repository, for compilation in isolation. So, the GMPlib search fix should occur as part of zarith (its build system is custom and isn't yet attempting to use autoconf/automake or something similar to avoid problems).

I added big integer support to the Erlang External Term Format implementation used by this source code when OCaml still included big integer support with num-1.1, due to big integer support being a part of OCaml at that time. I could remove big integer support if you think that is a good approach to resolve the problem (by removing the use of zarith).

I would like to think that zarith could be a part of OCaml in the future, but the current OCaml direction appears to consider big integer use as unimportant for OCaml use.

[mseri]

Thanks!

[mseri]

Perhaps I was too fast here at merging. What is the point of requiring zarith but also vendoring it?

[okeuday]

@mseri The approach is intentional to catch problems early and improve the in-place compilation in the main repository (https://github.com/CloudI/CloudI/tree/develop/src/api/ocaml) as necessary. Using this approach, there is always the assurance that the build of the OCaml CloudI API is the same as in the main repository.

[mseri]

In this case you don't really need this line

[okeuday]

While it might not be strictly necessary, I wanted to be transparent about the dependency on zarith, so it appears in the cloudi package information.

[mseri]

Thanks for the clarification