probe(audience): revert log-pane suppression to test Unity 6 stall hypothesis#767
Closed
ImmutableJeffrey wants to merge 3 commits intochore/audience-trim-unused-depsfrom
Closed
Conversation
…tries The previous matrix combined a cartesian product (unity x target x backend) with partial-key include entries that attached changeset and runner. Verified on runs 25604434621 (PR 754) and 25613048568 (PR 765) that the partial-include merge silently fails: matrix.runner ends up unset on every combo, runs-on resolves to empty, and the playmode job spawns zero jobs. Win/Mac PlayMode coverage has therefore been missing from every PR stacked on this branch. Replace with 12 fully-attributed explicit include entries, one per cell, plus the existing pull_request exclude that drops the 2022.3.62f2 cells on PR runs. This is a verbose but bulletproof shape: every key is set on every combo, no partial-key merge required.
…chAsync Temporary diagnostic for the Unity 6 Linux PlayMode regression. Tests on Unity 6 Mono Linux take ~33 sec each vs ~1.8 sec on Unity 2021.3 Mono Linux for the same suite, same llvmpipe, same network endpoint. T1 logs request start, T2 logs response received (continuation back on the main thread), T3 logs batch deletion done. Comparing T1->T2 vs T2->T3 gaps across Unity versions disambiguates whether the stall is inside HttpClient.SendAsync (network/socket/TLS regression in Unity 6 Mono on Linux) or in the awaiter continuation (SyncContext dispatch regression on Unity 6 Linux player loop). To revert: drop this commit.
…p llvmpipe rasterisation" This reverts commit f2f7037.
ImmutableJeffrey
force-pushed
the
chore/audience-revert-logpane-suppression-probe
branch
from
527720f to
58613ed
Compare
ImmutableJeffrey
changed the base branch from
chore/audience-linux-perf-test-polling
to
chore/audience-trim-unused-deps
Collaborator
Author
|
Probe complete. Hypothesis disproven: log-pane suppression was NOT the cause of the 30 sec/test stall on Unity 6 Linux. Removing it made Unity 6 cells slower by 5-7 min, confirming the suppression is a real win there. But the cause of the remaining 30 sec/test gap is elsewhere. PR #765 updated to keep the suppression but gate to Unity 6 only (Unity 2021.3 also got faster without it). |
ImmutableJeffrey
added a commit
that referenced
this pull request
May 10, 2026
Probe PR #767 showed Unity 2021.3 Linux cells regressed by 38s-2:24 with the suppression in place; Unity 6 Linux cells improved by ~7 min. Add UNITY_6000_0_OR_NEWER to the #if so 2021.3 and 2022.3 Linux cells are unaffected.
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.
Probe PR. Reverts f2f7037 only.
Hypothesis: the log-pane display:none on Unity 6 stalls UI Toolkit's panel scheduler, so SDK Log.Writer entries (queued via _logView.schedule.Execute) never reach the visual tree, and tests time out at the 30 sec WaitForLogEntry budget.
Evidence so far on PR #765 with the suppression in place:
If reverting the suppression brings Unity 6 cells close to Unity 2021.3 timings, hypothesis confirmed and the suppression approach was the wrong knob. If Unity 6 stays at ~30 sec/test, the cause is elsewhere and we look further.
Stack-style probe; not for merge.