Waleed Khan (@[email protected])

I'm a software engineer who builds highly-scalable developer tooling.

[Waleed Khan (@[email protected]) @arxanas]

New blog post: "Commit message test plans" — https://t.co/HsG5qj8bO8

[Waleed Khan (@[email protected]) @arxanas]

New* blog post: "🙏: please or thank you?" — https://t.co/wKTRTCcj8f

*Actually published for April Cools' 2026 https://t.co/WV59SrP2Lx

[Waleed Khan (@[email protected]) @arxanas]

New talk: "Datalog DSL detects defective dependency declarations, defanging dodgy development discipline" — https://t.co/uIv8xb5ers

Presented at the Seattle Build Enthusiasts Meetup (autumn 2025).

[Waleed Khan (@[email protected]) @arxanas]

New blog post: "Half my work is adding a cache" — https://t.co/US8DTIRNJY

[Waleed Khan (@[email protected]) @arxanas]

New blog post: "Lithe, less analysis with Datalog" — https://t.co/px0MMdQ2fB

[Waleed Khan (@[email protected]) @arxanas]

New talk: "Searching source control graphs" — https://t.co/s5kh2enzuP

Presented at the Seattle Build Enthusiasts Meetup (summer 2025).

[Waleed Khan (@[email protected]) @arxanas]

New* blog post: "Incremental processing with Watchman — don't daemonize that build!": https://t.co/oYhm532EW2

*Originally was a talk I gave at the Seattle Build Enthusiasts meetup on 2024-10-02, but I forgot to announce more widely

[Waleed Khan (@[email protected]) @arxanas]

New release: git-branchless v0.10.0 https://t.co/LX4BKQfXbe

[Waleed Khan (@[email protected]) @arxanas]

New blog post "Patch terminology" https://t.co/pygoVv06d3

[Waleed Khan (@[email protected]) @arxanas]

New blog post "Testing terminal user interface apps" https://t.co/fDd0fL6KTV

[Waleed Khan (@[email protected]) @arxanas]

@typesanitizer I vaguely recall @jordwalke saying that React was originally conceived in the context of SML (somehow using first-class continuations, possibly among other language features?), so it would very much be the encoding of a different architecture for the web (I could be mistaken).

[Waleed Khan (@[email protected]) @arxanas]

@typesanitizer Technically: the unidirectional data flow makes development more scalable and more "correct by construction" (compared to jQuery -> Backbone -> Angular — were you around for those?). Strategically: corporate backing and first mover advantage.

[Waleed Khan (@[email protected]) @arxanas]

@scott_klum Also check out https://t.co/9Dg3JcE0vE!

[Waleed Khan (@[email protected]) @arxanas]

@welltypedwitch Either 1) write a manual `fold` that doesn't return early, 2) use `fold_until` https://t.co/kzcDP6y6P1 or similar, 3) raise and catch an exception, or 4) suffer

[Waleed Khan (@[email protected]) @arxanas]

@_wilfredh I hacked my own GitHub comment support in the past, the main feature being that I wanted *inline* comments on the various paragraphs, rather than all comments being at the end (https://t.co/NlWdOnT5ob). I miss the commenting system from Real-World OCaml!

[Waleed Khan (@[email protected]) @arxanas]

@_wilfredh Where's the support for jj-style conflict markers 😤

[Waleed Khan (@[email protected]) @arxanas]

@Innf107 I see... % is line comment

[Waleed Khan (@[email protected]) @arxanas]

New blog post "Encoding ML-style modules in Rust" https://t.co/fgJVVqAIMc

[Waleed Khan (@[email protected]) @arxanas]

@_wilfredh I'd go farther and say that codegen should always succeed. It's not uncommon in a static language that's you want to make a change and try a certain code path, but are forbidden because the compiler wants you to fix the entire codebase first, when it could insert panics instead.

[Waleed Khan (@[email protected]) @arxanas]

@t3dotgg Imagine making `n` copies of your linked list and calling it good-looking