Revert "publish released mirrors' oleans to the Lake cache" - #10071
Open
kim-em wants to merge 1 commit into
Open
Revert "publish released mirrors' oleans to the Lake cache"#10071kim-em wants to merge 1 commit into
kim-em wants to merge 1 commit into
Conversation
Publishing worked: all 56 mirrors uploaded, the objects were publicly readable, and a consumer fetched every map and all 1128 artifacts they referenced. Consuming did not pay. Restoration is real but fails on exactly the expensive modules, so against a downloads-only cache HexGraphIso still recompiles 64 of its 91 modules while HexMatrix recompiles none. End to end the blog's CI took 41 minutes with the cache against about 31 without, measured twice with identical results. What those modules have in common is still unknown; it is not native_decide, a Mathlib revision difference, the upstream pins, precompileModules, or a root-versus-dependency effect, the last ruled out by a minimal reproducer. The credentials stay provisioned, so re-enabling this is a change to released-ci.yml alone once that question has an answer. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01BTzMFQfoJW6no7Zb8xwEMg
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
This PR removes the publish step from the mirrors' managed CI, reverting #10062 ("ci(release): publish released mirrors' oleans to the Lake cache"), and records what measuring it showed.
Publishing itself worked. All 56 mirrors uploaded, the objects were publicly readable, and a consumer fetched every map and all 1128 artifacts those maps referenced.
Consuming did not pay for itself. Restoration is real, but it fails on exactly the expensive modules. Against a downloads-only cache with every Hex build directory wiped,
HexMatrixrecompiles nothing in 2s andHexModArith2 modules of 96 in 43s, whileHexGraphIsostill recompiles 64 of its 91 in 367s andHexGF210. End to end the blog's CI took 41 minutes with the cache against about 31 without, measured twice with identical results, because the two to three minutes of fetching buys back less than it costs.What those modules have in common is still unknown. It is not
native_decide(one module in the whole graph uses it), not a Mathlib revision difference (the mirrors and the consumer pin the same one), not the upstream pins, and notprecompileModules. It is also not a root-versus-dependency effect: a minimal two-package reproducer, one published as root and consumed as a dependency, restores perfectly with and withoutprecompileModules, so cross-workspace reuse is supported.The credentials stay provisioned on every mirror, so re-enabling this is a change to
released-ci.ymlalone rather than a re-provisioning exercise.PLAN/Releases.mdkeeps the section, now describing the current state and pointing whoever picks this up at the open question rather than at the transport, which is sound.The reverted workflow has already been pushed to all 56 mirrors with the sync baseline advanced alongside, since the publishing tokens still lack the Workflows permission; a dry-run sync reports no divergence.
🤖 Generated with Claude Code
https://claude.ai/code/session_01BTzMFQfoJW6no7Zb8xwEMg