00:03 <d_bot> <mbacarella> is anyone aware of a PR in a project where ppx_tools_versioned was ported to use ppxlib? would help me out a lot as a teaching resource

00:08 <d_bot> <Anurag> This is a very small pull request so not sure if it’ll be helpful, but I moved a small library to ppxlib last year <https://github.com/hcarty/ppx_defer/pull/7/files>

00:45 <d_bot> <ansiwen> Ok, the server side is closing the connection after the error `[transport] transport: http2Server.HandleStreams failed to read frame: http2: frame too large`

02:20 <d_bot> <cod1r> hi

03:21 <d_bot> <mbacarella> nice. one question: did ppxlib require changing to use objects, or did you add that for funsies?

04:51 <d_bot> <sim642> Here's another example on a very small project: https://github.com/vogler/ppx_distr_guards/pull/1.

04:57 <d_bot> <sim642> I'm not too familiar with pre-ppxlib ways of doing these things, but looks like the old way to have an AST mapper was a record of functions which are called with the mapper record itself as argument, which is kind of a roundabout way to achieve open recursion. Ppxlib mappers are objects instead, where open recursion is directly a feature.

04:57 <d_bot> <sim642> Although I guess nobody's forcing those mappers to be used if you want to just perform direct recursion on the AST yourself.

06:20 <d_bot> <Bluddy> Using objects makes a lot of sense for something like ppxlib. We want to have open recursion, and we don't mind being slightly slower since it's not even part of the application.

07:01 <d_bot> <sim642> I wonder how these two approaches to open recursion compare performance wise. Has anyone benchmarked?

07:01 <d_bot> <sim642> I might soon have to do some refactoring to use open recursion in possibly performance-critical code, so it'd be interesting to know whether the extra boilerplate of the record-based approach would be worth it to gain a speedup over using objects directly.

07:09 <d_bot> <Bluddy> I started a discuss topic on this recently which didn't go that well https://discuss.ocaml.org/t/actual-performance-costs-of-oop-objects/9400

07:09 <d_bot> <Bluddy> my impression is that you take a serious performance hit, but so long as it's not in your critical path, it's probably ok

07:10 <d_bot> <Bluddy> records are much faster, but not every feature of objects can be emulated cleanly, and at a certain point you might as well just switch to objects

07:11 <d_bot> <Bluddy> also first-class modules are an option

11:27 <gopiandcode> heyo everyone!

11:28 <gopiandcode> Hacking on using the OCaml compiler as a library for my current project

11:29 <gopiandcode> Managed to get native compilation and execution of code working based of opttoploop

11:29 <gopiandcode> but only for a single module

11:30 <gopiandcode> when I have modules that depend on previously defined (and evaluated) modules, things crash

11:30 <gopiandcode> hmmmm....

11:30 <gopiandcode> anyone have any ideas what could be going wrong?

11:32 <gopiandcode> if it helps, the issue doesn't seem to be at runtime, but actually during compilation

11:32 <gopiandcode> when calling `Asmgen.compile_implementation`

11:35 <gopiandcode> the expception complains about unbound_var

11:38 <gopiandcode> https://github.com/ocaml/ocaml/blob/77b164c65e7bc8625d0bd79542781952afdd2373/asmcomp/selectgen.ml#L122

11:38 <gopiandcode> This is where the error is thrown from

11:42 <gopiandcode> Okay, looking into it, it seems like the env contains the binding for the module under it's actual name "Common", but for whatever reason, in the typed ast the module is represented as a variable "Common_178", causing the failure

11:44 <gopiandcode> maybe this is something to do with the coercions datatype returned by Includemod?

11:45 <gopiandcode> (`Includemod.signatures`)?

12:02 <gopiandcode> hmmm, looking more closely, it seems to be a discrepancy between the env returned by Typemod.type_structure and the typed structure

12:02 <gopiandcode> in the env, the module name is represented as Common

12:03 <gopiandcode> but in the typed structure item returned by Typemod.type_structure the module name is Common/178

12:03 <gopiandcode> this then gets compiled to a lamba expression of (let (Common/78 ...) ..),

12:04 <gopiandcode> when referring to common from another module, the lambda correctly refers to Common/78, but when passing it to Asmgen.compile_implementation, it tries to lookup Common_78 in env, and then crashes because it can't find it?

13:10 <olle> Damn, they used js_of_ocaml to compile the ReScript compiler to JS :)

13:10 <olle> https://cdn.rescript-lang.org/v9.1.2/compiler.js

13:10 <olle> Which means I can run ReScript + React in the browser, more or less

13:10 <olle> Have to try that later

23:03 <drakonis> how's the outlook on adding macros to ocaml?

23:17 <d_bot> <RegularSpatula> olle: why wouldn’t you just use ReScript-react if you want react with ReScript in the browser? (https://rescript-lang.org/docs/react/latest/introduction)