Florian Angeletti

The PR looks good to merge, thanks! I would propose to add a change entry in the manual section (something like "better integration with ocaml.org"?) to keep track of the...

You can add the change entry in the changelog in a new commit in this PR. When merging, I will squash the PR into one commit since this is mostly...

Since the regression is fixed in ppxlib, I am removing this PR from the 5.2 milestone. I am not sure if it still is worth it to discuss this PR...

I am not sure I follow why the alert printer would behave differently than the error printer?

They are the same printers? The `Astlib` module on the ppxlib side is including the compiler's `Location` module, and updating the shared `Location.input_lexbuf`.

This is an answer to a different question. To enable merlin support, it is sufficient to build `cmt` and `cmi` files for the compiler. Building the whole compiler is unnecessary.

Oops, I picked a non-representative subgraph, here is a graph of an unification step with the two arrows present.  The `,` nodes are indeed tuples. I agree that...

While squashing the history, I have added a basic description of the graph format; since I had been told that the format might be a bit too arcane to decipher...

This work has already upstreamed in #12451: we don't have an open world approach for builtin compiler attributes anymore. Instead, we record all known compilers attributes during parsing, and we...

The handling of module-level `since` tag is unfortunately a bit quirky in ocamldoc, that should be fixed once we switch to odoc for the documentation published on `ocaml.org`.