Repository navigation
Conversation
Tracked files named a private deployment host, its reverse-DNS launchd label, local binding names, and a home-directory path. Tests and docs now use neutral placeholders (pdpp.example.com, example-host, com.example.pdpp.*). The two canary manifests now carry the health probe as a path (/owner/login). The harness resolves it against PDPP_CANARY_ORIGIN, the same required, no-default variable the connector_run check already uses, and fails with a clear error when it is unset. Assisted-by: AI Signed-off-by: Tim Nunamaker <tnunamak@gmail.com>
Comments and test fixtures named a personal machine and a person. They now use neutral names (workstation, cin_vm, jane_doe_example_com). Test semantics are unchanged: one case where a user-chosen name shares no word with the connector title, one where it contains part of it. Assisted-by: AI Signed-off-by: Tim Nunamaker <tnunamak@gmail.com>
Tests and comments still named a personal machine, a person, a username, and a LAN address. They now use neutral values (workstation, Jane Doe, example-user, user@192.0.2.7 from the RFC 5737 range). Machine-name binding keys are renamed consistently within each test, and the example connection ids in one index test are fabricated. Assisted-by: AI Signed-off-by: Tim Nunamaker <tnunamak@gmail.com>
Tests and comments quoted real consumer email addresses and account handles from production observations. They now use fabricated values on example.com and neutral handles. Test shapes are unchanged: the label helpers and display-name rules do not depend on the mail provider domain. Assisted-by: AI Signed-off-by: Tim Nunamaker <tnunamak@gmail.com>
The binding rename changed the ids derived from it. The D8 and D9 Postgres tests now use ids and a legacy binding key recomputed with the real makeConnectorInstanceId, and both run green against a real Postgres. Production connection ids that appear only in comments next to renamed hosts and accounts are replaced with fabricated ones, and a work email in a consent-screen test is now on example.com. The canary now checks, right after the manifest parses, that PDPP_CANARY_ORIGIN is set whenever an http_health check uses a path. It exits before the build or any container action; before, a missing origin only surfaced after the deploy. The README states this. Assisted-by: AI Signed-off-by: Tim Nunamaker <tnunamak@gmail.com>
A set but malformed PDPP_CANARY_ORIGIN passed the preflight, then made resolveProbeUrl throw in the after-pass, which had no try/catch. The throw reached main's top-level catch: exit 1, no rollback, no receipt, new container left live. probeOriginProblem now also rejects an origin that is not an http: or https: URL. evaluateAfter turns a throwing probe into a failed outcome, so rollback and the receipt still run. One shared isHttpUrl helper replaces the two URL.canParse checks, so "localhost:8080/health" is treated as a path. Also removes a bare first name from a test comment. Assisted-by: AI Signed-off-by: Tim Nunamaker <tnunamak@gmail.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.
Tracked files named a private deployment host, personal machine names, people's names, account handles and real email addresses. They sat in test fixtures, code comments, a doc example and two canary manifests. This PR replaces each with a neutral placeholder:
pdpp.example.com,workstation,dev-laptop,Jane Doe,jane.doe@example.comand192.0.2.7(TEST-NET). Fixtures that need two distinct values keep two distinct placeholders, so test semantics are unchanged.The one behavior change is in the deploy canary. Its two manifests hard-coded the deployment URL in their
http_healthprobe, so they now carry the path/owner/login.resolveProbeUrlinreference-implementation/scripts/canary/deploy-canary.tsresolves that path againstPDPP_CANARY_ORIGIN, which theconnector_runcheck already requires. Absolute URLs pass through unchanged.records-keyset-index-and-bulk-lane.jsonhas noconnector_runcheck, so it now needs the origin where it did not before. A new preflight,probeOriginProblem, runs right after the manifest is parsed and exits with code 2 before the image build when a path probe has no origin or the origin is not an http(s) URL. Without it, the before-metric step would have logged the error and continued, and the after-check would have thrown after the container swap.evaluateAfternow also turns a throwing probe into a failed check, so rollback and the receipt still run. A new test incanary-harness-manifest.test.tschecks that a path resolves against the origin, that an absolute URL passes through, that a path with no origin or a malformed origin throws, and that a throwing after-check becomes a failed outcome. It failed on main because the export did not exist.Outside the canary script and the tests, code edits touch comments only. Every touched test file was run locally and none failed. Renaming a binding changes the connection ids derived from it, so the hardcoded ids in
device-enroll-postgres-admission-decoupling.test.tswere recomputed withmakeConnectorInstanceId, and that file was run against a real Postgres 16 (21 of 21 pass). Ids quoted only in comments were replaced with fabricated ones.device-exporter-storeandacknowledged-loss-store-postgresskipped their Postgres cases locally.Not covered here: #637 handles absolute paths in
reference-implementation/quality/**,PG-PROFILE-51-REPORT.md,consent-screen-client.tsx,explore-timeline-substrate.tsand the interop script, so this PR leaves those files alone. Opaquecin_ids in comments and fixtures are left as they are. Git history still holds the old values.Assisted-by: AI