Repository navigation
fix(release-summary): compare a stable tag with the previous stable tag - #326
Merged
Merged
Conversation
A stable tag is cut from its last rc's commit, so the tags-on-the-same- commit rule compared v4.12.2 with v4.12.2-rc.7. The range came back empty or near empty and collect.py skipped on zero merged PRs, so a stable banner in #product-releases got no useful summary. Denise asked for stable notes to cover everything since the previous stable tag, and pre-release notes to cover the previous pre-release, which they already do. A stable tag (no prerelease part, or patch / patch.N, matching linear-release-sync releaseIsStable from DEVOPS-1006) now compares with the highest earlier stable tag from the full tag list, since the previous stable can sit on another release branch out of reach of git describe (v5.0.0 on main follows v4.12.x on release-4.12). A -patch.N tag sorts right above the release it patches: in vcluster-pro v0.28.2-patch.1 is cut on top of v0.28.2, where semver order would put it below. Pre-releases and the previous-tag override keep their rules. COMMIT_LIMIT goes from 500 to 1000. A stable range spans its whole line: loft-enterprise v4.12.4 to the commit of v5.0.0-alpha.5 is already 716 first-parent commits, so at 500 the first v5.0.0 summary would drop its oldest work while the header said "since v4.12.4". Checked against loft-enterprise: v4.12.3 resolves to v4.12.2 (12 PRs, was v4.12.3-rc.2 with 5), v5.0.0-alpha.3 still resolves to v5.0.0-alpha.2 (39 PRs), and a v5.0.0 tag on alpha.5's commit resolves to v4.12.4 with 532 PRs, where main skipped with no commits. fit_budget drops every body for that range and still warns the input is over budget (464 KB against 400 KB); it proceeds, and stays inside the model context. Closes DEVOPS-1664 Claude-Session: https://claude.ai/code/session_017Fi5sA2RGopJjZnpWZfVxU
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.
Summary
The AI release summary for a stable tag compared against the rc on its own commit (v4.12.2 against v4.12.2-rc.7), so the range was empty or near empty and the summary skipped. A stable tag now compares with the previous stable tag; pre-releases keep comparing with the previous pre-release.
Key Changes
collect.py:is_stablematches linear-release-syncreleaseIsStable(no prerelease part, orpatch/patch.N). For a stable version with noprevious-tag,resolve_previous_tagreturns the highest earlier stable tag from the full tag list, so a previous stable on another release branch counts (v5.0.0 on main against v4.12.x).-patch.Ntag sorts right above the release it patches. In vcluster-pro, v0.28.2-patch.1 is cut on top of v0.28.2; plain semver order would put it below and compare it with v0.28.1. This deviates from the issue's "usingsemver_key" wording on purpose.COMMIT_LIMIT500 to 1000. loft-enterprise v4.12.4 to the commit of v5.0.0-alpha.5 is already 716 first-parent commits, so at 500 the first v5.0.0 summary would silently drop its oldest work.-patch.N, override, first-stable and a main() run of a stable cut from its last rc.action.ymlprevious-tagdescription and README state both rules.Verification
collect.pyagainst a full loft-enterprise checkout, no Slack, no AI:For the simulated v5.0.0,
fit_budgetdrops every PR body and still warns the input is over budget (464 KB against 400 KB). It proceeds; the output side (max-tokens: 16000) was not exercised without an Anthropic key.make test-release-summary(228 pytest, 13 bats) andmake lintpass.Dependencies
None.
TODO
release-summary/v1to the merge commit.release_summaryinrelease.yamland theprevious_tagdescription inrelease-summary.yaml(follow-up PR, References DEVOPS-1664).Closes DEVOPS-1664
https://claude.ai/code/session_017Fi5sA2RGopJjZnpWZfVxU