07:01 <reynir> @Swerve when you edit a message on discord it re-sends the whole thing on IRC (this channel is bridge with IRC and discord)

07:35 <discocaml> <NULL> Oh cool ! Less so if the message is huge, but at least changes aren't lost

08:05 <discocaml> <KW78> Interesting question. Is immutability enough to enable the zero cost aspect of it (which was the question initially)? Persistency is about sharing (transforming and potentially keeping previous version - of the tree here) so I though it was better to put it that way. If you google it you get nice images with "persistent data structure".

08:51 <discocaml> <octachron> Immutable data structures are tautologically persistent: persistent data structure are observably immutable data structures that may use inner mutability for efficiency reason.

09:25 <discocaml> <KW78> tks

14:07 <discocaml> <masterbuilder> I'm curious, is `ocamlgraph` generally considered to be a well-architected/idiomatic library? I just tried it out for the first time and while the library is very impressive and relatively easy to use, there's something about it that feels somewhat awkward. It seems excessively functor-heavy, requiring you to make new modules to be able to perform operations on your graph, which looks to me like it would tend to make the code very nounif

14:22 <discocaml> <bnguyenv> I think the interface is pretty dated, and I'm not aware of other libraries that are similarly functor-heavy. But it's a solid library.

14:34 <discocaml> <octachron> Mirage is even more functor-heavy.

14:34 <discocaml> <Bluddy> My biggest issue is locating the functions you want. The modules are a mess.

14:35 <discocaml> <Bluddy> The functors are there to allow supporting different variations of graphs.

14:50 <neiluj> hey! How do you force the evaluation in the Lwt monad? to get a value of state Return 'a

14:51 <Fardale> With Lwt.bind?

14:52 <neiluj> ah right! forgot to say that once the promise is resolved, I'd like to retrieve the returned value out of the monad

14:53 <neiluj> it is for a PBT test which expects the test code to be of the form unit -> bool

14:55 <neiluj> so maybe match Lwt.bind (f args) (fun res -> res) with Result r -> ... | _ -> assert false?

14:57 <octachron> You should rather put your tests in the monad. Then you can run the Lwt scheduler to run your test.

14:58 <octachron> (aka only the full program should bridge the monad with the external world)

15:00 <discocaml> <bnguyenv> Haven't used Mirage, but for `ocamlgraph`, I kind of feel like the functors could be avoided. Maybe with witness types like for `base`'s sets and maps ?

15:06 <neiluj> thanks octachron! the tests are of the form : make : name:string -> f:(t -> bool) and f uses something that returns in Lwt's monad

15:07 <neiluj> so build f prior to make in Lwt's monad would work?

15:12 <neiluj> hmm f is a combinator..

15:13 <octachron> I am not sure what you mean, but if `make` is scheduling tests by itsef, you need to write a version that is compatible with a `bool Lwt.t` return.

15:14 <octachron> and returns a test result int the `Lwt` monad.

15:17 <neiluj> oh ok this makes sense, thanks!