migrate zig v0.15.x > v0.16.0#41
Conversation
|
@chaploud, just a few conflicts and no update to the flake, yet. do you think you can work with this? |
|
Hi @notxorand, thank you very much for taking the time to put this migration together — it's genuinely appreciated. Heads-up on ordering: we have another PR (#28, "Support explicit cancellation of Wasm execution") that touches Once #28 is in, we'll circle back to this PR. Thanks again for kicking things off on 0.16 — we'll follow up here with next steps. |
|
Hi @notxorand — thank you so much for starting this migration work and opening this PR. It gave the effort a concrete shape early on, and the approach here informed a lot of what ended up landing. Because the Zig 0.16.0 migration turned out to touch quite a few additional surfaces (new Final result in #45: all CI green on macOS / Ubuntu x86_64 / Windows x86_64 (62,263 spec / 796 E2E / 46 real-world compat / FFI + static link on all three), Zig 0.16.0 toolchain, Thanks again — and please feel free to contribute to any of the follow-ups (e.g. the Zig-philosophy-aligned un-link-libc migration tracked as W46). |
Zig 0.15.2 → 0.16.0 toolchain migration, shipping as v1.10.0. Platforms: macOS aarch64 + Ubuntu x86_64 + Windows x86_64 all CI-green (62,263 spec / 796 E2E / 46 real-world compat / FFI + static link + minimal build). Notable surprises resolved in this PR: - Windows `std.c.fd_t == HANDLE` vs MSVCRT `_write(int fd, …)` ABI mismatch → WASI fd I/O routed through Win32 `WriteFile`/`ReadFile`/`SetFilePointerEx`. - `link_libc = true` required on Linux/Mac for WASI (std.posix lost `fsync`, `mkdirat`, `unlinkat`, `renameat`, `dup`, `pread`/`pwrite`, `futimens`, …). Linux ELF stripped binary grew ~290 KB, size guard 1.50 → 1.80 MB. - `std.posix.timespec` is `void` on Windows — cancellation tests use a comptime-branched Sleep helper. Deferred to follow-ups: - W46: Un-link libc (migrate to `std.Io` / `std.posix.system`). Returns Linux binary to 1.50 MB and aligns with 0.16's stated direction. - W47: `tgo_strops_cached` +24% regression on Mac aarch64 (20 other benchmarks improved 10-76% — GC paths notably faster). Thanks to @notxorand (#41) for opening the migration effort.
…-migration Empty commit crediting @notxorand as co-author (#41 superseded by #45). No code changes, no CI gate needed.
|
Hi @notxorand — one more follow-up, this time an apology. When #45 landed and shipped as v1.10.0, your contribution did not make it onto the repository's GitHub Contributors graph. The right thing to do at merge time would have been either to cherry-pick your commit ( To correct it without rewriting the v1.10.0 tag history (which downstream projects such as ClojureWasm pin to), I've landed a small empty commit on You should now appear on https://github.com/clojurewasm/zwasm/graphs/contributors — GitHub can take a few minutes to refresh the graph after a merge. Thanks again for opening #41. The 0.16 work started here, even if the final migration ended up taking a different path. |
…rewasm#41) PR clojurewasm#41 was the first attempt at migrating zwasm to Zig 0.16.0, opened by @notxorand. The final 0.16.0 migration shipped in v1.10.0 via clojurewasm#45 which took a different path, so clojurewasm#41 was closed without merge and @notxorand's commit (a0e049b) was not cherry-picked nor referenced via a Co-authored-by trailer on clojurewasm#45. This empty commit retroactively credits @notxorand so their contribution appears on the repository's GitHub Contributors graph (https://github.com/clojurewasm/zwasm/graphs/contributors) without rewriting the v1.10.0 tag history (which downstream projects pin to). Co-authored-by: notxorand <66304707+notxorand@users.noreply.github.com>
2 participants
No description provided.