Recruiters: Please put the name of the company in your initial pitch message. Don't make me message you back to ask for it.

"A company based in TOWN" is super unhelpful when you get a buch of those messages AND a lot of them are the same damn role. 😑

It finally got me on Sunday it was brutal. A bunch of staff and students out at my kids school and my work. Be safe out there. 😅

I seriously spent a LOT of time getting the playground in that state and it feels great being able to use this tool for something useful. Even if it was months later that initial investment was worth it (outside of just learning).

I just used the language playground I built for a different Lox implementation to disassemble some Lox code to make sure my new C implementation is compiling Lox correctly.

That's pretty fucking cool.

kyleect.github.io/locks/#/?cod...

I don't usually use snapshots in tests but I have been using them extensively in language implementation tests. It's really nice to create `name_of_test.ext` along with:

- name_of_test.ext.out - Expected stdout
- name_of_test.ext.err - Expected stderr
- name_of_test.ext.exit - Expected exit code

It's been a long time since I've felt the need to out time on my calendar to just work... Half my week has been meetings 😭

First my kid then my wife. I'm terrified to catch whatever this stomach bug is 😵‍💫

I have a bash script that looks for the *.ext source files then runs the language's CLI against them to compare against these snapshot files.

Honestly sometimes I'll add a source file test, update the snapshot and see what it outputs. If it looks good, commit it otherwise fix the implementation.

I don't usually use snapshots in tests but I have been using them extensively in language implementation tests. It's really nice to create `name_of_test.ext` along with:

- name_of_test.ext.out - Expected stdout
- name_of_test.ext.err - Expected stderr
- name_of_test.ext.exit - Expected exit code

That being said so far I actually dislike the single pass compiler.

Everything is so tightly coupled and I have no idea how I'm going to write tooling (LSP is a big one).

In a single pass tokens are directly translated to bytecode and that intermediate info is just lost.

I'm working on implementing clox (from the excellent Crafting Interpreters book) and it's a single pass compiler. This is neat because all the other languages I've worked on generated tokens, an AST, then bytecode.

So it's big that isn't fixed later in the tutorial but the bug also appears to be fairly minor 🤷‍♂️ Hey though, writing food tests actually caught that the actual behavior didn't match expectations.

Generating software faster, or automating busywork, doesn't guarantee success in business.

I also find short bursts of working on different code also works but not really viable while I'm at work.

Playing short bursts of chess or solving chess puzzles do wonders for raging (but medicated) ADHD. Like a grounding exercise.

TFW you're following a tutorial and go off the beaten path. You write some tests and discover something odd...

Is it a bug? Will it be fixed later in the tutorial? Did I mistype something?

Another thing I've noticed learning C is how fast compiling the project is.

Now this project isn't huge yet but less than a second is stunningly fast compared to my larger Rust projects creeping up in compile times. I say this with Rust being my daily driver language.

Time to finally take care of these flaky tests, eh?

I've had basic a gcc command up until this point but it's gotten annoying enough that I have a branch trying CMake. I've always heard horror stories and while it does look horror-ish, I can see the utility here.

What's funny is I'm adding another layer to this monstrosity by having just run it...

Ok, I have to say this... Aligning switch and case on the same indentation level hurts my brain.

switch (expr) {
case value:
// ...
break;
}

An excerpt from "Crafting Interpreters" showing how "embedding" a struct as the first field you can simulate inheritance by taking advantage of how structs are laid out in memory

I think this is the first bit of brilliance I've seen in C.

This bit (no pun intended) really made C click for me. The power of memory control outside of manual memory management. 🤯

I see Bluesky still has a bug where it drops you in the home feed if you change apps, even if you were writing a post. That's ok, I suppose I didn't need to post that right now 🙃

So there's no way to search for your old posts on LinkedIn? I have to be missing something here...

It's amazing how awful Google search results are now. I've posted this before but each time the results are actually worse than the times before.

I'm not using CMake and hand wiring all the files for compile/linking. For testing I'd obvious like the build to happen then the tests. I'm just not sure what's recommended in C land.

I've got the initial VM, a few OP codes, and a stack. What test tools do folks recommend for C? For languages I've found most of my tests are E2E where I'm running the language's CLI or the function that the CLI calls then asserting against stdout/stderr.

Any C tools for testing like this?

I'm revisiting Crafting Interpreters and implementing the clox language. This is my first serious dive in to actually writing C. I've been reading up on C as part of my interest in implementing languages but also while I was first learning Rust.

I have thoughts but I'm reserving them for now. 😅

Applicable to software