Repository navigation
docs(spike): direct batchexecute read latency vs the browser read (#1225) - #1232
Merged
Merged
Conversation
) N=10 per arm on denon82. Cached direct vs warm page at p50: as29s 2.6x, Zzl0ze 3.4x, nzlxg 1.8x - under the pre-registered 5x, so the direct tier is not justified by speed against a warm page. Against a one-shot cold Chrome it is 6-7x to the reply, and the cold client's graceful close hit its 30 s bound in 29/30 cycles (one real gflow credits call: 40.4 s wall). Part of #1229.
…ne database Ubuntu 24.04's tzdata 2026e moves backward links (Atlantic/Faeroe, Portugal, Eire, Iceland, Poland) into tzdata-legacy, which the CI runners do not install. ZoneInfo then refuses the alias, consent_region() fails closed, and the Atlantic/Faeroe row (expected False) went red on every PR, develop included; the three legacy rows expecting True passed for the wrong reason. The mapping test now stubs the zone lookup, so every row checks the name-to-region mapping on any machine. A separate test pins the fail-closed behaviour for an alias the installed zone database cannot load. (cherry picked from commit 9f356c9)
…1225) The verdict rests on the 6-7x cold-to-reply ratio; the ~31 s graceful-close timeout (30/31 clients) is a separate finding. Cold nzlxg is n=9 (the dropped sample was a >=30 s timeout) plus one 6.98 s top-up run alone.
|
Warning Review limit reachedYou've used all free OSS reviews for now. Wait for the free limit to reset to keep reviewing this public repository. Next included review available in 21 minutes. View limit details
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. |
This was referenced Nov 10, 2026
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.
Closes #1225
Part of #1229
Assessment (read-only, before the spike): a measurement question, not a bug report. No CLI or MCP surface changes. It is verifiable end to end here: $0, a headed browser on
denon82, reads only.Verdict (pre-registered: cached direct vs warm page, at p50; 5x = worth it, under 2x = not for speed)
p50 wall times, in seconds:
as29sstatusZzl0zelistingnzlxgbalancenzlxg(1.8x) is not worth a direct tier for speed alone.as29sandZzl0ze(2.6x and 3.4x) are worth it only with another reason.N = 10 per arm, so p95 is the maximum sample; the p95 column is in the doc. One cold
nzlxgsample failed with a 30 s or longer timeout and was dropped; one top-up run alone took 6.98 s.Separate finding (likely a defect, worth its own issue)
Gflow's browser close times out. In 30 of 31 clients the graceful
context.close()hit its 30 s bound (_engine._CONTEXT_CLOSE_TIMEOUT_S) before the force-close ran. One realgflow credits user --profile denon82took 40.4 s of wall time, about 31 s of it closing Chrome.Not measured
attoken when SNlM0e is missing from the HTML? #1221, PR docs(spike): source of the batchexecute at token (#1221) #1231).as29s.docs/FLOW_WIRE_RPCS.mdis not ondevelopyet (PR #1219), and this spike measures time, not wire shape, so there is no row update.Also carries
9f356c9d(cherry-picked from the open PR #1215, a test-only CI tz fix).Gates (local)
pytest -n 8 --dist worksteal: 6064 passed, 7 skipped.tests/scripts(404 passed) were re-run.