This sort of lead into trouble at one time, as the author chose to use the OCaml serialization of data as the protocol, so synchronization between 32- and 64-bit platforms or even binaries compiled with different versions of OCaml was not possible. Eventually this was fixed, though, with custom serialization.
Tho initially all those years back IIRC I wanted to tinker with how it installs mods for the Linux native "enhanced" editions. Otherwise I wouldn't even know what it's written in.
I've personally fixed an early noughties Perl script where one of the modules it used had changed its name (but thankfully not the interface, not significantly at least), among a couple of other compat problems. Fixing it turned out pretty straightforward even for someone who's far from a greybeard Perl "hacker".
A big part of the problem wasn't that you couldn't write good code with Perl, but that there were too many ways to write bad code with it. That made it anathema for any sort of corporate development, which is the main driver of programming language progress.
Personally, I'm glad that this isn't yet another Rust post ;)
Unlike Perl, OCaml is AOT compiled in a very efficient machine code, has a good static type system and has a good concurrency support. Both are not very mainstream.