Repository navigation
fix: scanner connection ID check, Close paths, EDS limits, deterministic allocation budgets (v0.7.1) - #29
Merged
Conversation
…, fix Close paths (v0.6.1) Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Codecov Report❌ Patch coverage is
📢 Thoughts on this report? Let us know! |
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
A patch release on v0.7.0. No exported identifier is added, removed or changed.
What changes:
ErrProtocoland closed at the target, as for the actual packet intervals v0.7.0 checks.OnStatemay callIO.CloseandScanner.Close, with Reconnect too; its goroutine and re-entry contract is documented. Reopen delays have equal jitter. AClosewhileOpenwaits for its reply leaves nothing running; of concurrentIO.Closecalls, one closes.Conn.Closefails the request in flight withErrClosedat once; a request whose write fails returns within its context; requests in flight duringClient.Closefail withErrClosed, as documented; a dialer that returns no connection and no error fails withErrConnectFailed.ServeCIP,Attribute.Set,ServiceFuncandSymbolHandleris documented.Parsebounds sections (16384) and fields in the file (1048576).Tests: the L1–L16 lifecycle matrix in TESTING.md with tests for concurrent Close, Close during a read or a write, callbacks that call methods and Close, and Open racing Close; goroutine leak checks in client, server, implicit and scanner; allocation-budget tests for the cip, encap and implicit decoders, eds.Parse and the server's handling of replies beyond their budget. These measure per call over many calls, the way
testing.AllocsPerRuncounts, with no other goroutine of the module running, against budgets derived from what the code must hold; this makesTestCPFAllocationBudget, which failed on ubuntu-24.04 in #29 with one measured call, deterministic. Fuzz targets for the Forward_Open and Forward_Close replies and connection failures;make fuzzdiscovers its targets; a nightly GOARCH=386 job;BenchmarkConsumemeasures delivery to the callback, with a new figure in TESTING.md.Documents: RELEASE.md (v0.7.1), TESTING.md, SECURITY.md, docs/LIMITATIONS.md, docs/IMPLICIT.md, docs/EDS.md, ARCHITECTURE.md, docs/FEATURE-STATUS.md, docs/api.
Not in this change: recovering handler failures at the library boundary (a behaviour change, for a minor release).
This pull request was first opened for a v0.6.1; v0.7.0 (#28) was released in the meantime, so it is now merged onto v0.7.0, keeps v0.7.0's implementation wherever both fixed the same thing, and targets v0.7.1.
Checked independently:
make check, otfabric-policy compliant, gorelease (v0.7.1 valid against v0.7.0), the regression tests fail on the v0.7.0 code, linux/386 test run.🤖 Generated with Claude Code