Elizabeth's work on Whamm triggered a New Stack article!

So that "simple" register allocator bugfix metastasized into a rewrite of about one third--approaching one half, TBH--of the graph coloring allocator. And I just realized that deconstructing SSA before doing the pre-spill (shadow stack) phase was...unwise.

Getting dragged down into the register allocator hole again. Every time I disappear down the hole I spend several days paging stuff back in, learning two or three minor insights, fix a bug by turning something off, and then promptly forget it all.

Ugh, I am so close to being able to squashing the last -O3 bug in the Virgil compiler! It passes all of Virgil's, with bootstrap, and passes nearly all of Wizard's, except some mysterious coverage monitor bug and WASI fd_write. After -O3 it's time to stable rev!

@shriram.bsky.social You always refer to your offspring as "kid". I wrote a song about the first Scientist. Somehow "Kid A" reminds of your stories of them. suno.com/s/07w8E55Fo9...

I submitted a PR to add Wizard to the features page for the official WebAssembly site: webassembly.org/features/ (cached preview doesn't seem to show that new column, though)

Haven't tried myself yet, but this VSCode extension for customisable debug visualisations looks absolutely amazing. marketplace.visualstudio.com/items?itemNa...

new bloggery: a look back on the last couple years of developments in v8's garbage collector wingolog.org/archives/202...

Today I presented to the Wasm CG a proposal for fine-grained dynamic code generation as a core WebAssembly feature. The proposal is now at phase 1! github.com/WebAssembly/... Also immortalized in song: suno.com/song/19e0679...

It’s live! "@graalvm.org meets #WebAssembly” from @devoxx.com: compile JVM apps to Wasm (Web Image) and run Wasm in Java or Kotlin at near‑native speed (GraalWasm). Lots of live demos included! www.youtube.com/watch?v=uefc...

How I'm feeling about data layout / protocol description languages right now. ...except none of them even ending up becoming standard! xkcd.com/927/

@dubroy.com alerted me to the Safari release notes (developer.apple.com/documentatio...) say version 26 ships an in-place interpreter for WebAssembly, which is in part based on the Wizard design, but adapted for Safari's use cases. This is cool! Wasm brings all the VMs to tiers!

Today I finished implementing the last of the relaxed-simd proposal (now Phase 4). That completes Wizard's support for all of the Wasm 3.0 features, which includes all the good things like exception handling, function references, garbage collection, and tail calls! github.com/titzer/wizar...

Hello, it's 2011 again and I'm crashing the MacOS kernel with 300 byte binaries. If only it were someone *else's* kernel.

Whenever a new feature is hot off the presses I am a little hesitant to use it in "production" code, and certainly a self-hosted compiler requires a stable rev to be usable, but man--unittests? Blast away with those lambda bombs. I have so much pent-up closuring!

That about does it for lambdas in Virgil. Woohoo! First there were just Class, Functions, Tuples, and Type Parameters. Then there were algebraic data types. Then, partial application. Now, full lexical closures! github.com/titzer/virgi...

As of last December, release 2.0 of the #Wasm specification is “official”! Read Andreas Rossberg's post, which happens to be the first real post on the #WebAssembly website's new News 🗞️ section: webassembly.org/news/2025-03....

Yesterday we put up the camera-ready for the WALI paper! This paper introduces Thin Kernel Interfaces for WebAssembly which allows a new class of powerful applications for Wasm and building higher-level interfaces like WASI in an engine-agnostic way. arxiv.org/abs/2312.03858

Wizard has supported Wasm GC for a while now, but now it finally supports all GC opcodes in JIT-only mode. The object model needs work, but this unblocks more perf work for stack switching. github.com/titzer/wizar...

Over the past week or so I spent some time adding some syntactic sugar (with type inference) to Virgil. The idea is that function/method bodies that are a simple return of an expression can have their return types inferred. A can of worms, but looks nice, IMHO. github.com/titzer/wizar...

Windows 95 was launched 30 years ago. I had a programming book on Windows 95...in 1994!! 500+ pages that documented darn near everything. Glossary, index, working examples, migration path. The amount of preparation that went into this launch was astounding. www.youtube.com/watch?v=_Jzf...

Finally pulled the trigger on another stable revision of the Virgil compiler (checking in the binaries of the compiler having compiled itself), which is good because right after is the perfect time to shake things up with new features and optimizations! github.com/titzer/virgi...

Last year I asked a question about the state of tracing JITs, and it led to a wonderful exchange. @cfbolz.bsky.social has written a terrific summary that captures a lot of folk knowledge that would otherwise be lost. Thanks! pypy.org/posts/2025/0...

Today I got to do some thinking about compiler IRs and finally solved a problem that has been vexing me for at least a dozen years: Is it possible to directly interpret a sea-of-nodes graph efficiently (i.e. no schedule)? What does that even mean? Yes Virginia, it's possible. It means max ILP.