Repository navigation
test: close the test and wording points that the reviews left open - #904
Conversation
…t open - TestResolveServeConfigEmptySyncAllowPrivate: serve reads an OPENSTORAGE_SYNC_ALLOW_PRIVATE that is present but empty as unset, the one layer that can tell it from an absent one (review of #897). - The docs and comments call OPENSTORAGE_SYNC_PUSH_ALLOW_PRIVATE the deprecated push-only variable, not a spelling: configuration.md also calls it a separate variable, not a legacy alias (review of #897). - TestReadValueListsPastGitArgBudget runs each missing-repository case as a subtest named after its route, and its comment says that only the v1 routes check the list first (review of #898). - TestSyncValuesTheRunRefusesAtRegistration gives each POST case a repository name of its own, and runs the tracked and repeat creates as a subtest, so one failure no longer hides the others (review of #900). - The stored-text row of wire-compatibility.md scopes the line-break token rule to a sync create (review of #900). Signed-off-by: Thomas Kosiewski <tk@coder.com> --- _Generated with [`xum`](https://github.com/coder/xum) • Model: `anthropic:claude-opus-5-5` • Thinking: `xhigh`_ Change-Id: Ifbee7306f2c6802dfae82bfab4ea5e110cfa4291 Signed-off-by: Thomas Kosiewski <tk@coder.com>
Main gained #896 and #901. The only conflict is the table row of OPENSTORAGE_SYNC_PUSH_ALLOW_PRIVATE: main names the metadata endpoints, this branch calls the variable a variable; the row keeps both. Signed-off-by: Thomas Kosiewski <tk@coder.com> --- _Generated with [`xum`](https://github.com/coder/xum) • Model: `anthropic:claude-opus-5-5` • Thinking: `xhigh`_ Change-Id: I0b86a3e1b2048dc56b084e500e59e21d06c6ddc7 Signed-off-by: Thomas Kosiewski <tk@coder.com>
- deploy-production.md: the strict webhook policy also denies a cloud metadata endpoint that is a public address (Azure's 168.63.129.16). - configuration.md: the webhook `false` bullet no longer files the metadata endpoints under the IANA special-purpose ranges. - ADR-45: most metadata endpoints lie in the private classes, not all. - guard.go: the note on 168.63.129.16 no longer reads as past tense. - The rest of the code comments, ADR-44 and the boot warning test call OPENSTORAGE_SYNC_PUSH_ALLOW_PRIVATE a variable, not a spelling. Signed-off-by: Thomas Kosiewski <tk@coder.com> --- _Generated with [`xum`](https://github.com/coder/xum) • Model: `anthropic:claude-opus-5-5` • Thinking: `xhigh`_ Change-Id: I8e3b4beb6d4674c53f67f9500ce58b6a2c49d3f0 Signed-off-by: Thomas Kosiewski <tk@coder.com>
Change-Id: I0d4451f09f2f4493c9a9743abff43828811feed9 Signed-off-by: Thomas Kosiewski <tk@coder.com>
…op test The review of #902 noted that the test of wcCheckRefPolicy wrapped the recorder in an envelopeWriter with no statuses, so it would keep passing if a commit write route later answered 403 in its result envelope. It now runs once for each route in commitWriteEnvelopeStatuses. The errtext README lists the unknown-op callers and the cut parse error of CheckRegistration. Signed-off-by: Thomas Kosiewski <tk@coder.com> --- _Generated with [`xum`](https://github.com/coder/xum) • Model: `anthropic:claude-opus-5-5` • Thinking: `xhigh`_ Change-Id: I65a72822779feee84988fa6b8544a60a2aa509af Signed-off-by: Thomas Kosiewski <tk@coder.com>
|
@codex review |
|
@codex security review |
Codex Review Summary: ✅ CompletedThis comment shows the latest Codex review activity on this pull request.
ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings. |
|
Codex Review: Didn't find any major issues. You're on a roll. Reviewed commit: ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
If Codex has suggestions, it will comment; otherwise it will react with 👍. Codex can also answer questions or update the PR. Try commenting "@codex address that feedback". |
🛡️ Codex Security ReviewSecurity review completed. No security issues were found in this pull request. Reviewed commit: Only the user who started this review can view the report in Codex. ℹ️ About Codex security reviews in GitHubThis is an experimental Codex feature. Security reviews are triggered when:
Once complete, Codex will leave suggestions, or a comment if no findings are found. |
Closes the P3 points that the reviews of #897, #898, #900, #901 and #902 left open. No change in behavior: tests, docs and code comments only.
OPENSTORAGE_SYNC_ALLOW_PRIVATEthroughserve's config:resolveSyncEgressgets a value-only lookup, so an empty value and an absent one are the same input thereTestResolveServeConfigEmptySyncAllowPrivate(cmd/openstorage/config_test.go) runsresolveServeConfigwith the variable empty (with and without the deprecated variable) and withfalseagainst a deprecatedtrue. A mutant that reads an empty value asfalseat the call site fails itOPENSTORAGE_SYNC_PUSH_ALLOW_PRIVATEa "spelling" in some places and "a separate variable, not a legacy alias" in anotherdocs/reference/configuration.md(table row, section intro), ADR-44, ADR-45,cmd/openstorage/README.md, and the comments incmd/openstorage/config.go,wire.go,internal/sync/egress.go,worker.goand the boot warning test call it the deprecated push-only variableTestReadValueListsPastGitArgBudgetnamed no route in its failure, and its comment held only for the v1 routesmissing repository/diff, …); the comment says that the path-style alias looks the repository up firstTestSyncValuesTheRunRefusesAtRegistrationused one repository name for every POST case, so a case that stored a repository turned the next ones into repeat creates; a failure in its tracked-create part skipped the credential casest.Errorfdocs/explanation/wire-compatibility.mdsaid "abase_repo.auth.token" for the line-break rule, which only a sync create appliesbase_repo.auth.tokenof a sync create"docs/how-to/deploy-production.mdsaid the strict webhook policy fails closed unless an address is public and outside the IANA registries, but a public metadata endpoint (168.63.129.16) is denied too; the webhookfalsebullet ofconfiguration.mdfiled the metadata endpoints under the IANA ranges; ADR-45 said the private classes hold every metadata endpoint; theguard.gonote on168.63.129.16read as past tensewcCheckRefPolicywrapped the recorder with no envelope statuses, so it would keep passing if a commit write route started to answer 403 in its result envelope;internal/errtext/README.mddid not list the unknown-op callerscommitWriteEnvelopeStatuses; the README listswrefPolicyCheck,wcCheckRefPolicy,wmCheckRefPolicyand the cut parse error ofCheckRegistrationTwo points of these reviews were fixed in their own PRs instead: the denied-address list of the sync
truevalue (#901) and the "any length" comment on zones (#903).Evidence
go test ./cmd/openstorage/passes (60 s).go test -race -tags=openstorage_viewcheckof./internal/events/...(22 s) and./internal/sync/...(75 s) passes at 469bed4, and so does./internal/api/...without pg (406 s). PostgreSQL and S3-WAL rely on CI.worker.gochange is a comment).Review
OPENSTORAGE_SYNC_ALLOW_PRIVATEread asfalseatresolveServeConfig; a 403 inrestore-commit's envelope statuses; the blame list check after the repository lookup, which fails onlymissing repository/blame range; the reserved-prefix check dropped, which reports both creates and still runs the credential cases). The two merges of main kept everything.Desk follow-up to the reviews of #897, #898, #900, #901 and #902.
Generated with
xum• Model:anthropic:claude-opus-5-5• Thinking:xhigh