#vscode and LLDB in VSCode were SO amazing to use. Within a couple of iterations, I was able to track down the use-after-free: https://github.com/artichoke/artichoke/pull/674 #fail #win #artichoke

After 50 PRs, I was able to put GH-442 to bed and remove the Rc wrapper from the Artichoke state https://github.com/artichoke/artichoke/pull/670 #win This refactor took 4 months #fail #rust #artichoke

It mostly works and is reasonably speedy, although while iterating, a template error in the lifesteam post rendering code caused ~1000 templates to fail compilation. I let the build peg the CPU for 20 minutes before giving up #fail

I made the mistake of exporting the database first AND not documenting how I transformed it to YAML, so I've been hesitant to post here #fail

hyperbola was down for 26 hours because the CA cert bundle on its backing RDS instance was not rotated before the 2015 bundle expired #fail #outage #aws

Nothing like a little build breakage to start the weekend https://github.com/artichoke/artichoke/pull/417 #artichoke #fail

Reported a critical bug in rustfmt #fail #github #patch https://github.com/rust-lang/rustfmt/issues/3770

Also added many more converter implementations with macros. There are 988 TryConvert implementations. #win #fail #artichoke

Led to reporting this #mruby bug: https://github.com/mruby/mruby/issues/4684 #github #fail although I was able to work around it #artichoke

The converters are central to #artichoke. Changing this core abstraction was painful: 84 files changed, 2000 lines added, 2800 lines removed. #fail #git

This was a nasty PR https://github.com/artichoke/artichoke/pull/242 #artichoke #fail

I'm trying to improve Regexp performance in #artichoke by using the regex crate instead of oniguruma in some cases. It turns out not to be faster in all cases #patch #rust #fail https://github.com/rust-lang/regex/issues/604

This was hard to track down: emscripten linker error with a particular function signature #rust #wasm #patch #github #fail https://github.com/emscripten-core/emscripten/issues/9135

I forked onig and hope to add support for the wasm-unknown-unknown build target. For #artichoke though, it's too soon. Once mruby-sys is pulled into #Rust I'll try again #ruby #mruby #fail #patch #Wasm

Neither mruby-sys nor onig can build with wasm-unknown-unknown or wasm-wasi, so I was stuck with wasm-unknown-emscripten. The linker on mruby-sys would dead code eliminate artichoke_backend::Artichoke #fail #Wasm #artichoke #mruby

With renewed focus of building my own #Ruby instead of extending #mruby, I put in some effort to make mruby an implementation detail of #artichoke. Lots of refactoring with sed. #fail

CactusRef is still an incredibly unsafe crate, but at least it aborts if it detects a use-after-free #fail #artichoke

That memory leak turned out to be quite the yak shave. setjmp/longjmp from C leaving #rust memory in inconsistent state #fail https://github.com/lopopolo/ferrocarril/pull/168

#git bisect is why it's important that your repo be buildable at every commit. Currently fixing a massive memory leak in ferrocarril #fail

Despite all I talk about #aws #ec2, I cannot for the life of me type isntances correctly #fail