Repository navigation
mcpx: conformance test package does not compile — test backend Complete returns []string, mcpserver.Backend wants Completion #3710
Description
Problem
pkgs/mcpx/internal/conformance (test package) does not compile against the current mcpserver.Backend interface: go vet fails with
internal/conformance/features_test.go:308:24: cannot use failingReads{…} (value of struct type failingReads) as mcpserver.Backend value in argument to mcpserver.New: failingReads does not implement mcpserver.Backend (wrong type for method Complete)
have Complete(context.Context, "encoding/json".RawMessage) ([]string, error)
want Complete(context.Context, "encoding/json".RawMessage) (mcpserver.Completion, error)
Evidence (2026-11-01, queue worker lane B, task 5601)
- The mismatch exists at origin/main itself, not just on a PR branch:
internal/conformance/harness_test.go:276definesfunc (b *backend) Complete(...) ([]string, error)andfailingReadsembeds that backend, whileinternal/mcpserver'sBackend.Completereturns(Completion, error)since 6cba232 ("mcpx: preserve completion total through the MCP backend (mcpx(daemon): /v1/complete never applies the 100-value spec cap — local fallback honors an unclamped completion.maxValues (0 = unlimited), upstream replies pass through verbatim #2878, Codex P2 on mcpx: shared 100-value completion cap on both daemon /v1/complete paths (#2878) #3505)"). - Observed failing PR mcpx: give tasks honest cancellation and error endings (#2499) #3699's CI (run 41310850598):
basic: lint(go vet) andstress: PR changed tests(darwin compile gate, same package) both fail on this; the PR does not touch the conformance harness or the Backend interface.
Acceptance
go vet ./pkgs/mcpx/...(or the CI lint job) passes with the conformance test package compiling: the test backend'sCompletereturnsmcpserver.Completion(preserving the completion-total semantics mcpx(daemon): /v1/complete never applies the 100-value spec cap — local fallback honors an unclamped completion.maxValues (0 = unlimited), upstream replies pass through verbatim #2878 introduced).- The conformance suite runs again in the stress/darwin compile gate.
Acceptance criteria
go vet ./pkgs/mcpx/...(or the CI lint job) passes with the conformance test package compiling: the test backend'sCompletereturnsmcpserver.Completion, preserving the completion-total semantics mcpx(daemon): /v1/complete never applies the 100-value spec cap — local fallback honors an unclamped completion.maxValues (0 = unlimited), upstream replies pass through verbatim #2878 introduced.- The conformance suite runs again in the stress/darwin compile gate.
- No production (non-test) behaviour changes unless the issue discussion records why one was necessary.
Metadata
Metadata
Assignees
Labels
Package/app: mcpx (pkgs/ or apps/ in this repo)Small: expected under 2 hours of workSomething isn't working (prefixed form of the classic bug label)Intake review done: acceptance criteria, links, and normalized labels addedImportant and time-sensitiveFiled or worked by the queue/worker pilotQueue packet kind: verify, reproduce, and route this item
Activity
developing-today commentedon Nov 1, 2026
CODE-1510
dezren39 commentedon Nov 1, 2026
Intake review
Related: PR #3699 (CI run 41310850598 showed the failure), PR #3505 / #2878 (commit 6cba232 changed
Backend.Completeto returnCompletion). No merged PR referencing #3710 was found at intake.Notes: The issue already carried an
## Acceptancesection (restated above as Acceptance criteria) and file-level evidence: the mismatch is on origin/main itself (internal/conformance/harness_test.go), not a PR branch. priority/high because it breaks the lint and darwin compile gates for unrelated PRs.Risks: Low: test-harness-only fix. The one trap is 'fixing' it by weakening the Completion total semantics #2878 added.
Suggested approach: Update the conformance test backend (and
failingReadsembedding) to the currentBackendsignature and let CI verify.🤖 Agent note · filed
developing-today commentedon Nov 3, 2026
agent-note:v1
state: reviewed (coverage loop-back)
Coverage pass 2026-11-03, loop-back review: this issue was re-reviewed (drift / overlap / already-handled / stale / blockers / ordering) so every open issue in the four repos is reviewed and in the system. Verdict: still live; in the system as task 7401 (triage, queued, priority 50) in repo-health-code-pilot. Labels audited and synced. If it is already handled at current main, its triage closes it with evidence; if oversized, the triage files the split; blockers belong on this issue.
Found via: coverage loop-back 2026-11-03
dezren39 commentedon Nov 4, 2026
🤖 Agent note · picked up
dezren39 commentedon Nov 4, 2026
🤖 Agent note · done
Triage verdict — already fixed (task 7401)
Verified fresh this shift against
origin/mainat3adad8fc9(merge of PR #5212):pkgs/mcpx/internal/conformance/harness_test.go:276now readsfunc (b *backend) Complete(_ context.Context, params json.RawMessage) (mcpserver.Completion, error)— exactly the signature the issue asked for, preserving the completion-total semantics from mcpx(daemon): /v1/complete never applies the 100-value spec cap — local fallback honors an unclamped completion.maxValues (0 = unlimited), upstream replies pass through verbatim #2878.pkgs/mcpx/internal/mcpserver/server.go:67, isComplete(ctx context.Context, params json.RawMessage) (Completion, error)— match.failingReads(the type named in the original vet error) embeds*backend(features_test.go:423) and is passed tomcpserver.Newatfeatures_test.go:308— it compiles through the correctedComplete.2f62ce6c9, 2026-11-02) with the corrected signature already in place — i.e. the fix arrived after the issue was filed on 2026-11-01, through the tree restructure, not through a PR referencing this issue.go vet ./internal/conformance/...— pass (no findings);go vet ./...frompkgs/mcpx— pass (no findings). That is the issue's acceptance command (go vet ./pkgs/mcpx/...) with the conformance package compiling.No production (non-test) behaviour change was needed or made. Closing issue #3710 as completed/already-fixed.