Gabriel Radanne
Gabriel Radanne
@lpw25 Do you still intent to change this ?
> So, the change is essentially blocked on having transparent module ascription. Fair enough. :)
I know this is WIP, but I'm curious. What would be the consequence for users, notably in term of FFI (and performances) ?
Hmm, maybe I misunderstood then ! If it's neither for FFI nor perf, what's the goal ? :)
@samoht Honestly, I don't remember. Maybe I just forgot to do it. Having eval/eval_cache separate was useful for handling failure, iirc.
At this point, it's pretty clear we want to keep the semantics of `with module` (where extra fields are inserted). The alternative semantics such as `with module M : S`...
I'm just going to answer the "eliom" part (the rest is great): No for now, maybe in a few month. Even if it's possible (currently, it's not), the build system...
During the functoria update, `mirage build` was only kept for backward compatibility reasons.
As far as functoria/the `Mirage` module/tool goes: > First choosing the storage backend (at compile-time), Yes, `generic_kv_ro` works like that. > and choosing the argument to the storage backend (ie...
Writing the documentation in `.mld` is a lot more future proof, given how `odoc` is moving along.