schmu VS ocaml

Since the last time I posted, I finished implementing pattern matching for schmu. To make matching on multiple columns less confusing I also added a tuple syntax to the language (finally), which are treated as anonymous records in codegen. Since then, I'm trying to overhaul my memory management, as my RAII-like solution only worked for linear code. In my first big departure from OCaml semantics, I decided to implement mutable value semantics. The paper linked in the Val language introduction makes a strong case for value semantics and after watching a couple of talks by Dave Abrahams, I wanted to try see how it feels. By making mutability be transitive and explicit, it also fixes one of the (few) gripes I have with OCaml that an array can never be really const as it is a reference type (it's possible to enforce constness with modules, but that's not exactly lightweight, syntax wise). Implementing mutable value semantics was pretty straight forward on the typing side, but I'm still not completely done with the codegen. This is due to 1. Assumptions about immutability I made in a lot of places are now wrong, and I had to completely change the way I pass values to functions. 2. I had to implement reference counted arrays, which was more work than I thought it would be. There are still edge-cases coming up in testing from time to time. Yesterday I finally managed it work for tail recursion, yay! I'm looking forward to getting rid of unneeded reference count updates in the future, by moving them to compile time, at least for linear code, lobster style. That's also an excuse to read that Perceus paper again. For the rest of November, I want to enhance my module system a bit. In particular, I want to add signatures and allow locally abstract types. I hope to have this in place before December to do the Advent of Code in my language.

I'm still working on my toy language schmu, an ML-inspired language which uses LLVM as backend.

I spent the time off over the Easter break to write the first program in my language which is not an explicit test and ended up implementing Ray Tracing In One Weekend. It was very rewarding to see how usable the language is already.

> OCaml’s configure script is also “normal”

If that’s this OCaml, it has a configure.ac file in the root directory, which looks suspicious for an Autotools-free package: https://github.com/ocaml/ocaml

You probably already know, but with OCaml 5 the only way to get flamegraphs working is to either:

* use framepointers [1]

* use LBR (but LBR has a limited depth, and may not work on on all CPUs, I'm assuming due to bugs in perf)

* implement some deep changes in how perf works to handle the 2 stacks in OCaml (I don't even know if this would be possible), or write/adapt some eBPF code to do it

OCaml 5 has a separate stack for OCaml code and C code, and although GDB can link them based on DWARF info, perf DWARF call-graphs cannot (https://github.com/ocaml/ocaml/issues/12563#issuecomment-193...)

If you need more evidence to keep it enabled in future releases, you can use OCaml 5 as an example (unfortunately there aren't many OCaml applications, so that may not carry too much weight on its own).

[1]: I haven't actually realised that Fedora39 has already enabled FP by default, nice! (I still do most of my day-to-day profiling on an ~CentOS 7 system with 'perf --call-graph dwarf', I was aware that there was a discussion to enable FP by default, but haven't noticed it has actually been done already)

11. OCaml - $91,026

> It partially helps since it forces you to have types where they matters most: exported functions

But the problém the OP has is not knowing the types when reading the source (in the .ml file).

> How would it feels like to use list if only https://github.com/ocaml/ocaml/blob/trunk/stdlib/list.ml was available,

If the signature where in the source file (which you can do in OCaml too), there would be no problem - which is what all the other (for some definition of "other") languages except C and C++ (even Fortran) do.

No, really, I can't see a single advantage of separate .mli files at all. The real problém is that the documentation is often worse too, as the .mli is autogenerated and documented afterwards - and now changes made later in the sources need to be documented in the mli too, so anything that doesn't change the type often gets lost. The same happens in C and C++ with header files.

If you have been in the Ruby community for the past couple of years, it's possible that you're not a super fan of types or that this concept never passed through your mind, and that's totally cool. I myself love the dynamic and meta-programming nature of Ruby, and honestly, by the time of this article's writing, we aren't on the level of OCaml for type checking and inference, but still, there are a couple of nice things that types with sorbet bring to the table:

Looks like there have been proposals to eliminate use of 3 operand lea in OCaml code (not accepted sadly):

https://github.com/ocaml/ocaml/pull/8531

An amazing example is Ocaml lang logo / mascot. It might be useful to talk with them to know what was the process behind this work. The About page camel head on Perl dot org header is also a pretty good example of simplification, but it's not a logo, just a friendly illustration, as the O'Reilly camel is. Another notable logo for this animal is the well known tobacco industry company, but don't get me started on that (“good” logo, though, if we look at the effectiveness of their marketing).

Haskell and Agda are probably the most obvious examples. Ocaml too, but it is much older, so its type system is not as categorical. There is also Idris, which is not as well-known but is very cool.

That does sound problematic, but without the code it is hard to tell what is the issue. Typically, compiling a 6kLoc file like https://github.com/ocaml/ocaml/blob/trunk/typing/typecore.ml takes 0.8 s on my machine.