Re: [MirageOS-devel] Using Result instead of Option in libraries

[Hannes Mehnert <hannes@xxxxxxxxxxx>]

On 14/10/2016 15:44, Anil Madhavapeddy wrote:
> On 14 Oct 2016, at 15:36, David Scott <scott.dj@xxxxxxxxx> wrote:
>>
>> Looks like we have a number of different conventions for the binds.  Do we 
>> want to have the same set of operators with and without Lwt support (and 
>> open the Infix module locally as needed) or separate operators that be used 
>> alongside each other?
>>
>> I'm curious what people recommend here -- I'm certainly open to adopting a 
>> common convention even if it involves a bit of churn.
>>
>> Initially I kept minting my own infix operators (like `>>*=` `>>|=`) but it 
>> quickly got confusing for me and I probably didn't name them uniformly 
>> across projects. Recently I've stuck to `>>=` and have put different ones in 
>> different modules like this:
>>
>> ```
>> module LwtResult = struct
>>   let (>>=) m f =
>> end
>> ```
>>
>> and now my code looks like (sometimes with extra newlines, but I don't have 
>> strong opinions about that)
>> ```
>> let open Lwt.Infix in
>> f () >>= fun () ->
>> let open LwtResult in
>> g () >>= fun () ->
>> Lwt.return (Result.Ok ()) (* possibly should define `return` in the module 
>> too *)
>> ```
>> Part of my rationale for using the same bind was a hope that one day modular 
>> implicits might automatically choose the right version and I could lose the 
>> `let open` but maybe this isn't possible?

I'd find the code without the `let open __ in` even more confusing (and
sometimes your `let open __ in` is already many lines away from its usage).

> Good point. To go over to the "overload the operators" camp despite my 
> previous mail, I did remember that ppx_let supports arbitrary monadic and 
> applicative syntax:

As said in my earlier mail, I find local `let open __ in` rather hard to
follow.

In the end, how many different binds do we need?  I'd think two or
three, the Lwt one (for effectful non-error-raising code), the rresult
one (for pure code), and the combined one >>=? : (a, b) result Lwt.t ->
(a -> (c, b) result Lwt.t) -> (c, b) result Lwt.t

And there will likely be combinations of >>= and >>=?

hannes

_______________________________________________
MirageOS-devel mailing list
MirageOS-devel@xxxxxxxxxxxxxxxxxxxx
https://lists.xenproject.org/cgi-bin/mailman/listinfo/mirageos-devel