diff options
| author | user <user@clank> | 2026-09-10 11:19:54 +0200 |
|---|---|---|
| committer | user <user@clank> | 2026-09-10 11:19:54 +0200 |
| commit | 16620e7fd9a6780756dc956466c3579e5bb2f384 (patch) | |
| tree | 4d7031ddfcf5b567accb41cabf4ffb1e8ed9d3ae /contrib | |
| parent | release: v0.1.7 paperwork -- formula and PKGBUILD pinned (diff) | |
| download | slosh-16620e7fd9a6780756dc956466c3579e5bb2f384.tar.gz slosh-16620e7fd9a6780756dc956466c3579e5bb2f384.tar.xz slosh-16620e7fd9a6780756dc956466c3579e5bb2f384.zip | |
build: a vendor bump rebuilds libghostty-vt, everywhere it is consumed
Three stale-build landmines, all found by shipping v0.1.7 -- the first
release where new code (GhosttySearch) actually needed the vendor bump
of 2026-09-01:
- $(VT_LIB) had no prerequisites, so an existing .a *was* the build:
the macOS builder compiled pane.c against zig-out headers from before
the bump. The vendor manifest is now the prerequisite; a re-vendor
rewrites it, so a bump rebuilds the lib and its installed headers.
- web/demo had the same guard and also never passed -Dversion-string:
upstream's build.zig ran git describe in our repo, and on the tag
checkout web/publish builds from, a slosh tag is a fatal version
mismatch. Both fixed; both had hidden behind a stale build/vt.
- The builder's zig-pkg (excluded from the sync, and left half-empty by
a once-interrupted push) is an input, not a cache: it is now synced
separately and additively, so a bump's new packages arrive and a
partial local tree cannot maim the builder's copy.
Diffstat (limited to 'contrib')
| -rwxr-xr-x | contrib/macos-remote-build | 12 |
1 files changed, 12 insertions, 0 deletions
diff --git a/contrib/macos-remote-build b/contrib/macos-remote-build index 55583ba..9442f93 100755 --- a/contrib/macos-remote-build +++ b/contrib/macos-remote-build @@ -152,6 +152,18 @@ else --exclude '/tests/fuzz/' \ --exclude '.DS_Store' \ "$root/" "$BUILDER:$DIR/" || fail "rsync failed" + # zig-pkg is excluded above (and protected from --delete) because it is + # 350M of vendored zig packages -- but it is an *input* to `make all`, not + # a cache: a vendor bump adds packages, and a builder missing one cannot + # build (v0.1.7 died on a translate_c that a once-interrupted sync had + # left as empty directories). So it is synced separately, additively: + # rsync sends only what changed, which after the first push is nothing, + # and no --delete means a partial local tree cannot maim the builder's. + if [ -d "$root/vendor/libghostty-vt/zig-pkg" ]; then + rsync -a "$root/vendor/libghostty-vt/zig-pkg/" \ + "$BUILDER:$DIR/vendor/libghostty-vt/zig-pkg/" \ + || fail "rsync of zig-pkg failed" + fi # web/ and tests/fuzz/ are excluded above but *tracked*, and .git went # over whole -- so the builder's stale copies of them read as local edits, # and `git describe --dirty` stamps a clean release tree -dirty (which is |
