task: first look at upstream v0.34.0, and the one divergence to carry - #290
Merged
Merged
Conversation
0.34.0 keeps the 0.33.3 native payload byte-identical (MLX, MLX-C, llama.cpp), so it is a Go/app release and every regression result in this fold's write-ups survives the sync. It also fixes the same runnerRef LogValue data race we fixed, four days earlier and independently, with TryLock instead of a second mutex -- which permanently drops four attrs from the 11 log sites that already hold refMu, and ships no test. Record the comparison and the decision to keep ours at the sync. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Written after checking 0.34.0 ahead of today's merges.
MLX_VERSION,MLX_C_VERSION,LLAMA_CPP_VERSIONbyte-identical), so 0.34.0 is a Go- and app-side release: every regression result in this fold's write-ups stays valid across the sync, and a Go-only binary swap measures it without a CUDA rebuild.app/,x/mlxrunner(xgrammar, speculate),cmd/,server/.x/mlxrunner/client.gogainsrequestGrammarand does not touch admission.x/models/nn/nn.go,x/mlxrunner/mlx/ops_extra.goand the generated bindings are untouched.runnerRef.LogValuerace on 2026-09-05 inb5d373f3(fix data races in progress and sched ollama/ollama#18319), four days before server: fix the data race between runnerRef.unload and LogValue #289 and independently, usingrefMu.TryLock()with the fields omitted when contended. Becausesync.Mutexis not reentrant, the 11 of ~25 log sites that already holdrefMualways fail that TryLock and permanently losename,inference,pidandnum_ctx; upstream also ships no regression test. The doc records the comparison and the decision to keep ours at the sync, resolve theLogValueconflict in our favour, and offer ours upstream so the divergence can be retired.🤖 Generated with Claude Code