Repository navigation
feat(desktop): package and provision independent Ops resources - #2503
Conversation
hidb4ai
left a comment
hidb4ai
left a comment
There was a problem hiding this comment.
The generated product resource configuration exceeds process environment limits at the actual helper inventory size, preventing the documented product build. Please address the inline finding before merging.
hidb4ai
left a comment
hidb4ai
left a comment
There was a problem hiding this comment.
Re-reviewed 96d6590. The previous P1 is fixed: the directory mapping keeps the configuration bounded, and the actual pinned Tauri CLI now reaches beforeBuildCommand for both the original counterexample and a full-size synthetic inventory. The macOS bundle regression also passed, including nested dotfiles. I found no new correctness defect.
The remaining merge gate is CI: Linux Desktop validation exceeded its 40-minute limit while installing OS dependencies, so its build/tests never ran. CI result consequently failed. This is an incomplete validation run, not evidence of a new source regression; please rerun the affected validation and obtain a green aggregate before merging.
knqiufan
left a comment
knqiufan
left a comment
There was a problem hiding this comment.
Reviewed this head: no new blocking findings. Focused resource tests passed; the real Tauri bundle test, Windows and installed-product qualification were not rerun in this review. GitHub does not allow the PR author to approve their own PR.
96d6590 to
143688d
Compare
hidb4ai
left a comment
hidb4ai
left a comment
There was a problem hiding this comment.
Re-reviewed 143688d after the rebase. The product-resource implementation and bounded Tauri configuration fix are unchanged; replaying the original and full-size inputs with the pinned CLI still reaches beforeBuildCommand. I found no new correctness defect.
Current-head validation is still incomplete: check-website was killed with exit 137 during page-data collection after compilation and TypeScript checks succeeded. The workflow was subsequently cancelled, leaving unit/e2e, upgrade and aggregate checks unfinished. These logs do not establish a new PR source regression. Please investigate the killed build and rerun the required checks for this head before merging.
143688d to
a365ba3
Compare
hireview001
left a comment
hireview001
left a comment
There was a problem hiding this comment.
- CI failed: CI result (failure).
- CI failed: harness-validation / Atomic Memory clean migration (OceanBase) (failure).
- CI failed: harness-validation / Acceptance (oceanbase) (failure).
- CI failed: harness-validation / Acceptance (sqlite) (failure).
- CI failed: skill-guidance / validate (failure).
- CI failed: harness-validation / Atomic Memory clean migration (OceanBase shard 0/3) (failure).
- CI failed: harness-validation / Atomic Memory clean migration (OceanBase shard 1/3) (failure).
- CI failed: harness-validation / Recall effort (OceanBase) (failure).
- CI failed: opendal-package (failure).
- CI failed: dsh-package (failure).
- Additional CI issues: PR checks.
a365ba3 to
917918c
Compare
hidb4ai
left a comment
hidb4ai
left a comment
There was a problem hiding this comment.
Re-reviewed abac4ba. I found no new correctness defect in the added OpenDAL, skill-guidance or DSH changes, and the original Tauri configuration counterexample remains fixed. The previous website CI failure is also cleared.
Required validation is still blocked: Python 3.11 unit tests fail in test_history_events_correlate_a_real_spawned_attempt_and_keep_queue_time_distinct while waiting 10 seconds for the spawned worker result. The same timeout reproduces locally on both this head and base 63f021d6, so this is not a PR-introduced finding. Other test jobs were cancelled, and the Upgrade release gate fails because required upgrade acceptance was skipped. Please resolve the test failure and obtain successful required checks for this head before merging.
Bring in the 6 first-parent commits that landed since the previous merge, including the OpenDAL fix for the Pydantic diagnostic that failed `opendal-package` on this branch and on master (oceanbase#2503), the Pi extension tests against a real Server (oceanbase#2531), and the harness workflow restructuring (oceanbase#2526). The merge is textually clean. The persistence digest in `release/upgrade-support.json` is unchanged. The new Pi service chain test passed the removed `scheduler_path` argument to `create_server_app`, so that argument is dropped. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Which issue or RFC does this PR close?
Related to #2112, #2108 and #2109. Platform acceptance and publication remain tracked there.
Rationale for this change
Product builds need to carry the approved independent Ops helper and bind their Server catalog to an approved distribution URL and the exact bundled qualification.
What changes are included in this PR?
bundlefallback for packages that omitops/.Are there any user-facing changes?
Packagers can embed the helper without requiring globally installed Python. Server installation and service mutations retain their explicit plan/approval boundary. Candidate packaging remains unchanged.
This does not publish or sign resources, qualify a new Server build, or provide installed-platform acceptance. Release owners approve download availability and immutability. Old and damaged helper copies are retained, and complete inventories are read on launch.
How was this change tested?
UV_OFFLINE=1 make checkandmake opendal-testpassed (10 OpenDAL tests).63f021d6;git diff --checkpassed.Cloud CI for this head. The local resource fixtures are synthetic; these checks do not establish Windows or installed product qualification.
AI usage statement
OpenAI Codex assisted with implementation, review, rebase and validation. Open Code Review delegate mode supplied file and rule selection.