Add ABI diff check for C API and C++ bindings - #387
Open
ethanglaser wants to merge 15 commits into
Open
ethanglaser wants to merge 15 commits into
ethanglaser wants to merge 15 commits into
Conversation
Merged
ahuber21
pushed a commit
that referenced
this pull request
Sep 14, 2026
Contributor
|
abicheck/integration-lab#2 - it's possible to use GH actions for some of this work an dthey can post messages to PRs in case of removals/additions |
napetrov
reviewed
Sep 14, 2026
Co-authored-by: Nikolay Petrov <nikolay.a.petrov@intel.com>
The shared library's header root is the whole of svs, which does not finish parsing in an hour; --depth binary compares it in seconds. Unused by this repo's own check, which only covers the runtime bindings, but abi-check.sh is the shared primitive and its env-override surface is its API -- the innersource nightly's release comparison is the consumer. Also treats abicheck's exit 5 (budget exceeded) as a finding rather than a harness bug, and retargets comments that described the deleted innersource-vs-public equivalence check. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Neither macro is referenced by any header the tarballs ship (grep over all 1992 headers of the runtime bindings tarball: 0 hits each; they are only read from .cpp files). Both build paths force IVF on -- innersource passes -DSVS_EXPERIMENTAL_ENABLE_IVF=YES and public -DSVS_RUNTIME_ENABLE_IVF=ON -- so the comment claiming the two sides disagree on SVS_ENABLE_IVF was also wrong. Verified the runtime comparison produces a byte-identical report without them. The svs/index/ivf/* header exclusions stay: svs/index/ivf/common.h really does include <mkl.h> unguarded and no tarball ships MKL headers. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Mirrors the runtime bindings' check: build-c-api-bindings.yml grows an abi-check job comparing this build's svs-c-api tarball against the newest downloadable one from a main push run, through the same .github/scripts/abi-check.sh the innersource repo calls, so neither repo can drift on how libsvs_c_api.so is measured. The C API is the best-conditioned target of the three libraries. It is compiled -fvisibility=hidden and exports only `svs_*` -- which the existing integration test independently asserts -- so there are no mangled names, no template instantiations leaked from header-only dependencies and no std:: internals in the exported set. svs_c.h reaches stdbool/stddef/stdint and its own generated version header, so full header depth is both affordable and meaningful. The suppression file is therefore empty, and stays a file rather than being omitted because abi-check.sh skips a missing --suppress path silently: a typo there would otherwise be indistinguishable from deliberate emptiness. No release-relative check, in this repo or innersource: no published release ships a C API, so there is nothing to compare against yet. The artifact's retention on main goes to 30 days, since it is now the baseline every open PR is measured against. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
ethanglaser
changed the title
ethanglaser
marked this pull request as ready for review
September 22, 2026 19:32
ethanglaser
requested review from
homksei,
mihaic and
yuejiaointel
as code owners
September 22, 2026 19:32
Member
Author
Pushed a branch off of this one that does this: https://github.com/intel/ScalableVectorSearch/compare/dev/eglaser-abi...dev/eglaser-abicheck-ghaction?expand=1, overall similar level of complexity to current state |
Member
Author
|
ABI check currently failing because available artifacts from last merged PR expired after 1 week. Fixing by rerunning those, but this PR would also extend that window to 30 days. |
This branch has not been deployed
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
No description provided.