Release Performance

release-performance.yml

427 workflow runs

chore(main): release 0.194.0 (#1335) * chore(main): release 0.194.0 * chore: refresh release lockfiles --------- Co-authored-by: chio-labs-delivery[bot] <320578061+chio-labs-delivery[bot]@users.noreply.github.com> Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
completed successfully:
Release Performance#352:Commit4d9e49cpushedbychio-labs-delivery[bot]Bot
main
Nov 7, 202628m 56s
feat(ui): wire the runs pages to the live control API (#1334) * fix: name no nodes in refused listings A refused location listing named the nodes the person could not view and read differently for a location that does not exist. Listing refusals now carry one fixed message with the decision ID, identical for unassigned and unknown locations. The audit event keeps the missing nodes. * fix: name no nodes in authorization denials Every refusal from person authorization now shows fixed text per action an...
completed successfully:
Release Performance#351:Commit68a1551pushedbykvlonge
main
Nov 7, 202629m 21s
feat(orchestration): let people read run nodes, triggers and ticks (#1332) * fix: name no nodes in refused listings A refused location listing named the nodes the person could not view and read differently for a location that does not exist. Listing refusals now carry one fixed message with the decision ID, identical for unassigned and unknown locations. The audit event keeps the missing nodes. * fix: name no nodes in authorization denials Every refusal from person authorization now shows fixed text ...
completed successfully:
Release Performance#350:Commitfb13201pushedbykvlonge
main
Nov 7, 202623m 33s
chore(main): release 0.193.0 (#1331) * chore(main): release 0.193.0 * chore: refresh release lockfiles --------- Co-authored-by: chio-labs-delivery[bot] <320578061+chio-labs-delivery[bot]@users.noreply.github.com> Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
completed successfully:
Release Performance#349:Commit8ba2f5apushedbychio-labs-delivery[bot]Bot
main
Nov 7, 202636m 35s
feat(cli): make sqlbuild a small forwarder to sqb (#1333) * feat: make sqlbuild a small forwarder to sqb sqlbuild was a second full copy of the sqb launcher. It now finds the sqb installed beside it and hands over the same arguments and streams: exec on Unix, a child that alone handles Ctrl+C and Ctrl+Break on Windows. Without sqb beside it, it names the missing program and exits 1. The bin compiles only the launcher hand-off file, so it stays at about 0.37 MB stripped instead of 6.4 MB when linking t...
completed successfully:
Release Performance#348:Commit821385dpushedbykvlonge
main
Nov 7, 202634m 41s
feat: serve the web UI from sqb server and ship it in the wheels (#1329) * feat(ui): add the orchestration UI prototype as a SvelteKit app Port the prototype pages, components and styles into ui/ as a static single-page app, with neutral example owners and data. * fix(ui): resolve the prototype's type-check errors Wire the lineage hover handlers to the current event names, declare the run gantt element refs as state and add Node types for the generated tsconfig. * feat(ui): add a typed API client la...
completed successfully:
Release Performance#347:Commit1345322pushedbykvlonge
main
Nov 6, 202633m 58s
feat: run the orchestration UI shell on the live control API (#1328) * feat(ui): add the orchestration UI prototype as a SvelteKit app Port the prototype pages, components and styles into ui/ as a static single-page app, with neutral example owners and data. * fix(ui): resolve the prototype's type-check errors Wire the lineage hover handlers to the current event names, declare the run gantt element refs as state and add Node types for the generated tsconfig. * feat(ui): add a typed API client layer ...
completed successfully:
Release Performance#346:Commitcf07c40pushedbykvlonge
main
Nov 6, 202633m 31s
fix: let Recall's final delivery rows use the remaining shutdown budget (#1325) * fix: let Recall's final delivery rows use the remaining shutdown budget The delivery summary rows written when a session closes were capped at 500 ms even when most of shutdown_timeout remained, so one slow hosted store response marked an otherwise fully delivered transcript incomplete. They now get the rest of the shutdown budget, and at least 500 ms; close still returns by the shutdown deadline. * test: pin the deadlin...
completed successfully:
Release Performance#345:Commit04e98cbpushedbykvlonge
main
Nov 6, 202634m 59s
ci: limit each timed dense compile to 4 GiB instead of the pytest tree (#1324) The dense guard job ran make, uv and pytest inside one 4 GiB systemd unit, so roughly 0.5 GiB of harness memory counted against a limit meant for the compiler. Each timed sqb compile now runs in its own transient scope with MemoryMax=4 GiB and no swap, and pytest runs outside it. Budgets are unchanged. Under CI the guards refuse to run without the per-compile limit, and a missing systemd-run or sudo fails naming the scope.
completed successfully:
Release Performance#344:Commitcd38ff9pushedbykvlonge
main
Nov 6, 202625m 34s
ci: limit each timed dense compile to 4 GiB instead of the pytest tree
completed successfully:
Release Performance#343:Pull request#1324synchronizebykvlonge
ci/dense-compile-memory-scope
Nov 6, 202628m 50s
fix: publish audits that share a name as separate snapshot nodes (#1326) Inline audits compile to their definition name, so two models with the same inline audit gave two snapshot nodes one key and publication failed. Attached audits are now keyed by their execution identity: audit name, target kind, target and column. Singular audits keep their key.
completed successfully:
Release Performance#342:Commit8f0728epushedbykvlonge
main
Nov 6, 202639m 42s
fix(security): name no nodes in authorization denials (#1322) * fix: name no nodes in refused listings A refused location listing named the nodes the person could not view and read differently for a location that does not exist. Listing refusals now carry one fixed message with the decision ID, identical for unassigned and unknown locations. The audit event keeps the missing nodes. * fix: name no nodes in authorization denials Every refusal from person authorization now shows fixed text per action an...
completed successfully:
Release Performance#341:Commit8d2d2dapushedbykvlonge
main
Nov 6, 202635m 27s
test: stop scheduler tests depending on runner speed (#1330) * test: stop scheduler tests depending on runner speed Four orchestration tests held a real 30-second scheduler lease across setup that is not about lease expiry, so a slow runner fenced them before the code under test ran. They now hold the lease for the whole test through a helper beside expire_lease, replacing per-chunk renewals. The platform webhook test waited a fixed ten seconds for 107 deliveries. It now fails only when the running di...
completed successfully:
Release Performance#340:Commitb858c24pushedbykvlonge
main
Nov 6, 202625m 16s
test: drain every due trigger after scheduler fault recovery (#1327) The fault test checked recovery with one scheduler cycle, which evaluates at most 100 due triggers. With 400 triggers, a slow runner that crossed a minute boundary during recovery left the rest without their latest tick. Run enough cycles to drain every trigger due at the settled time.
completed successfully:
Release Performance#339:Commit0455861pushedbykvlonge
main
Nov 6, 202631m 5s
chore(main): release 0.192.0 (#1323) * chore(main): release 0.192.0 * chore: refresh release lockfiles --------- Co-authored-by: chio-labs-delivery[bot] <320578061+chio-labs-delivery[bot]@users.noreply.github.com> Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
completed successfully:
Release Performance#338:Commita788b94pushedbychio-labs-delivery[bot]Bot
main
Nov 6, 202631m 18s
ci: limit each timed dense compile to 4 GiB instead of the pytest tree
completed successfully:
Release Performance#337:Pull request#1324openedbykvlonge
ci/dense-compile-memory-scope
Nov 6, 202634m 3s
feat: add microbatch coverage endpoint (#1319) * feat: read check evidence through the control API People with view.evidence on a check's model can read the sample rows stored with one of its results. Columns are the rows' keys in sorted order, the order the warehouse stores them in, and a row without a key holds null there. Every read is audited, and a read whose audit cannot be written discloses nothing. The SQL and evidence reads now share one authorize, read and audit path. * feat: read maintenanc...
completed successfully:
Release Performance#336:Commitd9abd83pushedbykvlonge
main
Nov 6, 202645m 48s
chore(main): release 0.191.0 (#1321) * chore(main): release 0.191.0 * chore: refresh release lockfiles --------- Co-authored-by: chio-labs-delivery[bot] <320578061+chio-labs-delivery[bot]@users.noreply.github.com> Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
completed successfully:
Release Performance#335:Commita99ff90pushedbychio-labs-delivery[bot]Bot
main
Nov 6, 202634m 25s
feat: add maintenance history endpoint (#1318) * feat: read check evidence through the control API People with view.evidence on a check's model can read the sample rows stored with one of its results. Columns are the rows' keys in sorted order, the order the warehouse stores them in, and a row without a key holds null there. Every read is audited, and a read whose audit cannot be written discloses nothing. The SQL and evidence reads now share one authorize, read and audit path. * feat: read maintenanc...
completed successfully:
Release Performance#334:Commitec9a3b4pushedbykvlonge
main
Nov 6, 202631m 25s
chore(main): release 0.190.0 (#1320) * chore(main): release 0.190.0 * chore: refresh release lockfiles --------- Co-authored-by: chio-labs-delivery[bot] <320578061+chio-labs-delivery[bot]@users.noreply.github.com> Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
completed successfully:
Release Performance#333:Commitbb756d6pushedbychio-labs-delivery[bot]Bot
main
Nov 6, 202623m 47s
feat: read check evidence through the control API (#1317) People with view.evidence on a check's model can read the sample rows stored with one of its results. Columns are the rows' keys in sorted order, the order the warehouse stores them in, and a row without a key holds null there. Every read is audited, and a read whose audit cannot be written discloses nothing. The SQL and evidence reads now share one authorize, read and audit path.
completed successfully:
Release Performance#332:Commitbf4a2a2pushedbykvlonge
main
Nov 6, 202626m 57s
chore(main): release 0.189.0 (#1316) * chore(main): release 0.189.0 * chore: refresh release lockfiles --------- Co-authored-by: chio-labs-delivery[bot] <320578061+chio-labs-delivery[bot]@users.noreply.github.com> Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
completed successfully:
Release Performance#331:Commit9ae0f5epushedbychio-labs-delivery[bot]Bot
main
Nov 6, 202630m 30s
feat!: read checks through the control API (#1315) People can list a target's checks with their newest results, read one check's recent results, and, with view.sql, read its SQL. Checks are keyed by binding_key, now a typed, indexed column on the audit-result mirror, because one audit name such as not_null covers many models and columns. A model rename or a severity or threshold change starts a new check history, and the old one stays readable. SQL reads are audited like node SQL. BREAKING CHANGE: cont...
completed successfully:
Release Performance#330:Commit865f8f7pushedbykvlonge
main
Nov 6, 202636m 16s
refactor: own local workers on every platform (#1313) Local worker ownership, recovery and stop now build on every platform. The CLI drops the Linux/macOS/Windows cfg gates and their fallback paths, and sqlbuild-process-ownership becomes an unconditional dependency. No behaviour change on Linux, macOS or Windows.
completed successfully:
Release Performance#329:Commit6726252pushedbykvlonge
main
Nov 6, 202631m 14s
ci: compare release performance with a release PyPI serves (#1314) Release Performance now compares against the newest release PyPI actually serves. A main commit uses its own release once published and the newest earlier one until then; a release commit never compares with itself. A missing or uninstallable baseline still fails the job.
completed successfully:
Release Performance#328:Commit7a2c531pushedbykvlonge
main
Nov 6, 202638m 13s