@joyeecheung.bsky.social
She/her. I contribute to Node.js and V8. I post about museums, food, plants and other stuff too. Blog: https://joyeecheung.github.io/blog/
@joyeecheung.bsky.socialWe are currently disabling metagen, it's just a matter of where the source of truth is for the time being, but when V8 stablizes it and the C++ part becomes the source of truth we will still need libclang issues.chromium.org/issues/53134...
@joyeecheung.bsky.socialWorking on upgrading V8 in Node.js to 15.5 and started Jenkins for the first time...and only the linter failed
@joyeecheung.bsky.socialThe cheese level is just out of this world
@joyeecheung.bsky.socialFood at @nodeconf.eu in Bologna is just another level 🥹
@joyeecheung.bsky.socialYes!
@joyeecheung.bsky.socialManaged to get my travel document back in time to make it to Bologna today…and sure enough they now check passports and visas even for flights from Spain 🤨
@joyeecheung.bsky.socialAnother thing I have been working on is to get it to be able to attach to a running process, log something, then detach when some specified condition is met. Should help with containing the output volume too, and avoid the restart/help hit difficult reproductions
@joyeecheung.bsky.social..or maybe through git blame, but the original spec is long and changes get merged from proposals wholesale. Another thing is that C++ spec anchors name a concept so it’s stable, whereas in ecma262 they are often named after some method used at the time and are less stable throughout refactoring
@joyeecheung.bsky.socialAlso for looking into how the C++ spec has evolved to reference a CWG issue, I could look it up in www.open-std.org/jtc1/sc22/wg... once I obtain the anchor. I am not sure how to do that for ecma262, maybe search in GitHub issues or across organization but they aren’t usually indexed with an anchor
@joyeecheung.bsky.social..whereas the ecma262 multipage is bigger, not paged by an easy anchor (sec-source-text-module-record-execute-module) and the anchor-as-id in the page doesn’t work in the web version actually
@joyeecheung.bsky.socialYes that’s what I also used, but for example compare:
And
The C++ multipage spec has smaller pages (therefore easier to consume and navigate), anchored directly by expr.static.cast and using the paragraph number it’s easy to locate.
@joyeecheung.bsky.socialObservations: - It was easier to find multipage spec and check resolution history with the C++ spec anchor compared to ecma262 - pre-commit CI coverage of these high-impact C++ compiler projects is really small when compared to Node.js/V8 - There is a visible test262-shaped void in the C++ world
@joyeecheung.bsky.socialThere were more docs to update when changing semantics in Clang, had to educate myself with the release process before finishing the commit
Clang commit
github.com/llvm/llvm-pr...
@joyeecheung.bsky.socialThere was a bit of a shock for me when navigating in the GCC codebase and the git blame annotation in my editor said the code was as old as myself.
GCC commit
gcc.gnu.org/cgit/gcc/com...
@joyeecheung.bsky.socialLanded fixes in GCC & Clang for [expr.static.cast]/8 wrong-code bugs when investigating a V8 bug related to it. Previously: enum Flag : bool { kNo, kYes }; static_cast<Flag>(2); // kNo in GCC ❌, kYes in Clang ✅ static_cast<Flag>(0.5); // kNo in GCC and Clang ❌
@joyeecheung.bsky.socialGem of my trip to Astorga: Palacio de Gaudí! Beautiful inside out and there is a very interesting epigraph museum in the basement 😇 Cathedral is great too, but if you put it next to Gaudí…😬
@joyeecheung.bsky.socialThis just landed! Thanks @addaleax.bsky.social and @legendecas.bsky.social for the reviews! It will be out in the next semver-minor release of 25, and likely backportable to older LTS - the new workflow is a compatible improvement to the existing postject-based SEA building workflows from v18.x.
@joyeecheung.bsky.socialFinished two retrospective blog posts on the journey of require(esm) before 2025 ends:
