Repository navigation
Sync repos to current template versions, with three-way merges - #33
Merged
Merged
Conversation
…y merges
`check` reported that a repo was behind; nothing could bring it back. Every repo
on python-service 1.0.0 has reported behind since 1.1.0 shipped, with no way up.
ansari sync payments --dry-run
ansari sync --fleet ~/src --pr
For each attached template that is behind, each file is handled by what its
recorded hash says: untouched files are replaced with the new output, edited ones
three-way merged, deleted ones left deleted. Files the new version adds are
written; files it drops are removed if untouched and kept (and untracked) if
edited. Overlapping edits get conflict markers and a non-zero exit.
A three-way merge needs what the old version generated, and this build only
ships each template's current version. The manifest recorded that output's
sha256, so sync finds it in the repo's own git history -- the committed blob with
that hash -- and merges with `git merge-file`. The new version renders with the
variables the repo was scaffolded with; new variables take their defaults, and
schema-1 string values are typed on the way.
A template is refused, with nothing written for it, when an edited file has no
ancestor in history, a new file would land on an untracked one, a destination
belongs to another template, or a newly required variable has no value. The
whole sync refuses a repo that isn't a git working tree, has uncommitted changes
(unless --allow-dirty), or carries a template this build can't resolve. Planning
does all the work, merges included, so --dry-run reports exactly what a real run
would.
--pr commits on ansari/sync-<template>-<version>, pushes, and opens a pull
request through the git and gh CLIs, which already hold the user's credentials,
so ANSARI holds no token. It opens nothing while any template is refused or
conflicted: conflict markers are never committed. --fleet syncs every repo under
a directory, one pull request each. This lives in integrations/, the only code
that touches the network; every command goes through a runner tests replace.
The acceptance test upgrades a real python-service 1.0.0 repo: the three 1.0.0
sources come from history into tests/fixtures, and a test confirms they reproduce
all seven files byte for byte against the golden hashes. After sync, the
deployment carries the autoscaling guard and values.yaml keeps its local edit.
Tests 236 -> 273 (scaffold + cli), 310 in the full suite; coverage 97%.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Marks v0.8 done and documents sync: what happens to each file, where the merge ancestor comes from, the refusals, and --pr. Corrects two statements sync made wrong: the README said the merge ancestor comes from re-rendering the old template (it comes from git history), and that the recorded variables exist to reconstruct the old version (they render the new one). Retires the "manual python-service 1.1.0 upgrade" limitation for the true remaining one, and records that the sync --pr tripwire can no longer be crossed. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
pr-reviewer-bots
Bot
added
documentation
|
Thanks for opening this PR. Assigned to @sameeralam3127. Labels added: |
4 tasks done
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
1 participant
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.
v0.8:
checkreported that a repo was behind; nothing could bring it back. Every repo on python-service 1.0.0 has reported behind since 1.1.0 shipped.On a real python-service 1.0.0 repo, with a committed local edit
How each file is handled
The merge ancestor comes from git history. This build only ships each template's current version, but the manifest recorded the sha256 of what the old version generated — so sync finds that exact blob in the repo's own history and runs
git merge-file.Refusals (nothing written)
--allow-dirty) · a template this build can't resolve--prand--fleet--prcommits onansari/sync-<template>-<version>, pushes, and opens a PR via thegitandghCLIs — they hold the credentials, ANSARI holds no token. No PR while anything is refused or conflicted: conflict markers are never committed.--fleetsyncs every repo under a directory, one PR each. All of it lives inintegrations/github.py, the only code that touches the network; tests swap the runner and never push.Acceptance test
Upgrades a real python-service 1.0.0 repo: the three 1.0.0 sources are pulled from history into
tests/fixtures/, and the test first proves they reproduce all 7 files byte for byte against the golden hashes.Gates
sync.py97%,github.py100%)ruff/format --check/mypy --strictDocs updated, including two README statements sync made wrong (where the merge ancestor comes from, and what recorded variables are for).
🤖 Generated with Claude Code