Merge from upstream - #8
Merged
Merged
Conversation
preview now carries the same tree as main and dev (36aed0b). The version string is the only difference, which is what the release workflow requires: preview publishes prerelease versions under the 'preview' dist-tag. Before this, the preview channel was 12,065 lines behind dev and still shipped the Compatibility Lab on every install's request path.
chore(preview): promote dev for the v2.20.0 line
[WRONG BRANCH] Merge dev into preview: Windows suite green
Merge dev into preview: ignore leftover test temp files
Merge main into preview: v2.24.2 release
Promote dev to preview: Wave 5 campaign (107 commits)
Promote dev to preview: CodeQL #87 ReDoS fix + closeout correction
Promote dev to preview: Wave 5 record corrections
Promote dev to preview: alert-precision record
Promote dev to preview: post-scan closing note
Promote dev to preview: final Wave 5 errata
Promote dev to preview: Wave 5 closing record
Promote dev to preview: v2.25.0 release train
release: v2.25.0-preview.20260818
promote: dev to preview for v2.26.0-preview.20260819
release: v2.26.0-preview.20260819
…window (#2526) #1905 taught catalog sync never to raise a compaction threshold retained from Codex. The rule is right, but the retained number was trusted even when sync corrected the row's context window in the same pass. An upstream entry arriving as 128k/115_200 whose window is then widened to 272k kept the stale 115_200 — 42% of the real window — so every long turn compacted early. CI caught it on macos and test 1/4 at 121c1fb. A retained threshold only describes the window it arrived with. Capture the incoming window before any override or cap rewrites the row, and trust the retained value only when the window is unchanged; lower-is-policy still holds there, which is what #1905 was protecting.
Docs-only roadmap for three operator-visibility defects that share one shape: OpenCodex computes the truth and does not report it. - 010 (#2457): the sidecar pair check collapses a five-member union into a two-arm ternary, so a submitted gemini backend is validated against the stored openai backend and the dashboard save 400s. - 020 (#2411): collectStatus already computes routingKind and ships it in status --json, but the human renderer never prints it, so a healthy proxy reads green while nothing routes through it. - 030 (#2412): a version-manager overwrite leaves a message-less ineligible verdict, and the CLI warns only when a message exists. 002 records the plan audit, including two amendments: WP4 must plumb messages to every reachable silent ineligible return rather than only the one the reporter hit, and WP3 must pin custom-local/unknown as intentionally silent.
…2492) * fix(gui): inset the Claude account-pool warning and threshold field `.card` carries no padding of its own — `.card-row`, `.card-sub` and `.setting-row` each supply the 16px inset. The experimental warning box and the threshold field are none of those, so in the Anthropic account-pool card they rendered flush against the card border while the title row and the rotation rows sat 16px in. Measured in the dashboard at 814px card width: title 17px from the card edge, warning box and threshold input 1px. The warning takes margin rather than padding: it draws its own border, so padding insets only its text and leaves the border on the card edge. The threshold field takes padding because the `.input` inside is full-width, so the field owns the inset for the label, the control and the help line. This is the same fix `.account-pool-strategy-card` already carries for the Codex pool card, applied to the card that still had the gap. Moving the box styles out of inline `style` is part of it: an inline `padding` outranks any stylesheet rule, so leaving it there would keep the inset unreachable. * fix(gui): keep the inset rationale in one place The stylesheet rule, the test file header, and three per-assertion comments each restated the same two facts — that `.card` has no padding and that the warning box needs margin rather than padding. CodeRabbit's slop detector flagged the repetition, and it was right: the explanation belongs on the CSS rule, where the values it justifies live. The test header now points at that rule and keeps only what is specific to the test — why the assertions read source text rather than measuring, and the browser measurements the contract is derived from. The comments that survive per assertion are the ones the CSS does not carry: why the rule regex is anchored, how a shorthand is read, and why an inline style would silently disable the fix.
fix(gui): show the device code and authorization link on every login surface
fix(gui): render the login hint during a first-time provider add
feat(oauth): let the operator decline a proxy-side browser open
fix(oauth): read code and state from a redirect URL fragment
fix(oauth): never pair a code and state from different URL components
merge dev into preview for the v2.33.0-preview.20260825 release
…parity fix(release): run the preflight suite in the same groups CI does
merge dev into preview for the v2.33.0-preview.20260825 release (release-gate parity)
…on-oracle test(usage): stop asserting the overlay version against a moving oracle
# Conflicts: # package.json
[WRONG BRANCH] merge dev into main for the v2.33.0 release
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
16 participants
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.
No description provided.