Gabriel Scherer

Fine with me! I do think we could make progress by merging what we already have, and then working on these as follow-ups rather than delaying the merge. (We could...

I tend to agree with @v-gb, in particular that having a dedicated literal syntax would be nice. Something that I think should be on the table is enforcing the existence...

There is also `c"a"` (equivalent to `'a'`) and then `uc"a"`, which as pointed by @v-gb offers much less surface for compatibility issues. I'm fine with any proposal that gathers consensus,...

(I briefly wondered if the type variable name `u'a'` could be rewritten `\#u'a'`, but I propose to just forget that I ever asked because it looks ugly and would require...

A remark on backward compatibility: - Dune generates its module-alias-introducing code itself during builds. This means that it can be adapted to a different syntax if we want -- and...

@lpw25 and @xavierleroy: I'd like to encourage @clementblaudeau to work on an implementation of this RFC (as part of his OCaml Foundation-funded stint on the module system implementation). Does that...

> consensus on the option name. My proposal: `-barrier-placement=[fence-before-store | branch-after-load]`.

To enforce compatibility between compilation options at runtime, I would put the information in the `.cmx` file (see file_formats/cmx_format.mli). I am not sure how to model architecture-specific linking options. Maybe...

We only keep bytecode compilers in boot/, so there should be no change necessary there. But yes, .cmx magic files would need an update.

This should have been fixed by #12029, so from a distance I don't understand why we have the problem again.