Upstreaming plan

[dbuenzli]

I have now added tests. In test.ml we ensure that all uchars codec according to spec and a few other tidbits. In test_long.ml I have a longer test (~8min on my machine) that ensures we do not decode invalid byte sequences to any uchar, I don't think people want that in the test suite…

In any case I think an upstreaming PR sequence could be:

1. Make a PR for Bytes and Uchar with the codec.
2. Make a PR for String.
3. Make a PR to make Buffer use the Bytes encoders.

The day Bigarray.Bytes exists I'll happily make a PR to add them there aswell ;-)

What do you think @gasche @nojb ?

P.S. I'll have to switch to something else in the upcoming days, I hope to be able to get back to it at latest by mid october.

[nojb]

Sounds good! Personally I think doing 1. and 2. together would be OK as the implementation for String should be a trivial wrapping around that of Bytes.

[dbuenzli]

Sounds good! Personally I think doing 1. and 2. together would be OK as the implementation for String should be a trivial wrapping around that of Bytes.

Yeah I just want to avoid c&p churn in case there is still discussion on names and doc strings. (Working on this already took me much more time than I expected…)

[dbuenzli]

In the end did 1. and 2. together which is now merged.

For reference here are all the PRs:

ocaml/ocaml#10710
ocaml/ocaml#10733