Cut the history out of CLAUDE.md - #22
Merged
Merged
Conversation
The file instructs; it is not a changelog. Six passages narrated what the repo used to do rather than what it does, and two of those had already outlived the state they described. Dropped outright: the claim that no -linux-<arch> tags remain in GHCR, which was true of the workflows and false of the registry; and the bullet restating that publishing is master-only, already said in full three bullets above. Merged: the pull-request validation bullet absorbed the one explaining why it works, which repeated the push-by-digest rule from the sub-bullet above it. Trimmed to the rule and its measurement: the esp-matter timeout no longer recounts the 180-minute figure and the 337-minute QEMU leg it was set against; the verification layer no longer explains that emulation used to be the loudest source of silent install failures; the cache bullet no longer says the settings "were removed", it says there are none.
There was a problem hiding this comment.
Pull request overview
This PR refines CLAUDE.md to be purely instructional (rules + rationale) rather than partially historical, removing outdated or redundant narration that could mislead readers about current CI/registry behavior.
Changes:
- Removed obsolete statements (e.g., about
-linux-<arch>tags and master-only publishing phrasing that duplicated existing rules). - Merged/condensed adjacent bullets to keep only the operative rule and still-relevant rationale (e.g., PR digest pushes leaving untagged blobs).
- Trimmed historical measurement backstory while preserving the measurements that still constrain current decisions (timeouts, verification layer rationale, build cache outcome).
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
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.
CLAUDE.mdinstructs; it is not a changelog. Six passages narrated what the repoused to do rather than what it does, and two had outlived the state they
described — the failure mode #16
was written to fix, arriving from the other direction: not a rule that drifted out
of step with CI, but a rule that was only ever a note about a migration.
Dropped outright
-linux-<arch>tags in GHCR any more." True of the workflowsand false of the registry, which held 51 of them until they were deleted today.
It read as a statement about what is published, so anyone opening GHCR and seeing
idf-v5.4.1-linux-amd64had to decide which source was lying.three bullets above, in the sub-bullet that already splits which tags get
written from whether the image is uploaded.
Merged
The pull-request validation bullet absorbed the "That works because…" bullet
following it, which restated the push-by-digest rule from the sub-bullet directly
above them both. The one fact only it carried — that a pull request leaves untagged
blobs behind — moved into the surviving bullet.
Trimmed to the rule and its measurement
337-minute QEMU leg it was set against. The measurement that constrains the
decision (8–11 minutes native) and the warning not to tighten toward it both stay.
source of silent install failures. The reasons that still apply — a stale mirror,
an installer exiting 0 on a partial install — stay.
removed", and absorbs the "cold by construction" clause that had been stranded on
the
workflow_dispatchbullet as a dangling reference to it.Every passage that justifies a rule with a number is untouched; only the narration
is gone. Net −31/+23 lines.