Run frontier MoE models on hardware you already own — pure C, zero deps, experts streamed from disk. Tiny engine, immense model. 🐦
C
ci(release): attach to an existing release instead of failing on it
JustVugg256b8fb795b6
diff --git a/.github/workflows/release.yml b/.github/workflows/release.ymlindex 11d166f7..b5b6b6a0 100644--- a/.github/workflows/release.yml+++ b/.github/workflows/release.yml@@ -210,6 +210,25 @@ jobs:if [ -z "$NOTES" ]; thenNOTES="Release ${TAG}"fi- gh release create "$TAG" artifacts/* \- --title "colibri ${TAG}" \- --notes "$NOTES"+ # Attach to an existing release rather than failing on it.+ #+ # `gh release create` errors with "a release with the same tag name+ # already exists" whenever the release was published by hand before+ # the tag build finished -- which is how any release with hand-written+ # notes happens. v1.5.0 hit exactly that: all three platforms built,+ # the checksums printed, and then the job went red having attached+ # nothing, so the release had notes and no downloads until the+ # artifacts were recovered from that run by hand.+ #+ # Uploading is the operation that matters and --clobber makes it+ # idempotent, so upload when the release exists and write notes only+ # when creating one. Notes a human already published are never+ # overwritten by the CHANGELOG fallback.+ if gh release view "$TAG" >/dev/null 2>&1; then+ echo "release $TAG already exists — uploading artifacts to it"+ gh release upload "$TAG" artifacts/* --clobber+ else+ gh release create "$TAG" artifacts/* \+ --title "colibri ${TAG}" \+ --notes "$NOTES"+ fi