rabbiveesh

The not having PHP locally installed is a reasonable decision - if your dev environment is dockerized, then you can debug it for everyone at the same time (I have wrangled many MANY local envs in my day, and I prefer it this way). Of course, for the same price you can enbug it - we had an issue that we changed the image for clickhouse when bitnami stopped hosting their image, and the new image was dumping all sorts of statistics and logs, and then eventually bloated >10GB of the stuff, leading to crashloops which ate up ALL of the virtual CPUs in the docker virtual machine. but i digress

Part of the reason i didn't just pick PHP also was that I was testing a workflow where I would let claude write the whole thing, while just making sure to validate it externally. My experience with agents writing PHP was less than stellar - there are a lot of sharp edges that humans don't realize. Rust is a lot more reliable when you need to change a bunch of things, cuz it simply won't compile (and helpfully so) until you get it right.

The experience where I was working was that nobody would run dump-autoload in their dev environment because it would take upwards of five minutes. This was even after solving all of the surprises that we had with our docker setup.

In our case, the docker issues were somewhat endemic, because different people would touch parts of the local setup, without realizing what downstream consequences there would be.

Part of the preference is being able to run on the host without installing PHP, so you don't pay the virtualization boundary tax (we were on Mac, so it's small but not nothing)

I built this a while ago b/c composer is written in php for (almost) no reason. Rust is faster, and it's also Rust. The goal was to cover the annoying bits of working on a php project - install and autoload generation. Install is parallelized via rayon, as is autoload generation. It's so fast that i stuck it in a git hook and stopped thinking about it. The former behavior was never run install or autoload because it was too slow (especially when docker is involved)

Thanks! I'm very actively developing it right now so if there are feature requests please open issues on github

I'm currently writing perl with a team, and I find that so long as you write in a way that looks good, there's no issue with other people reading the code either.

And that still happens at stream-of-conciousness.

were the answers at least in the form of questions?

I write new things in Perl all the time. Deploying highly functional backend services using Mojolicious, a framework which is VERY actively developed.