reflex-dev/reflex - CodSpeed

reflex

Blog Docs Changelog

Performance History

Latest Results

reflex-workflow: give the waiter-cap test a worker of its own It took session_factory and never used it, so the only thing settling the under-cap call was the worker that fixture keeps running -- an unstated dependency the sibling wake tests do not have. It runs an only_worker like they do. The assertion that the over-cap call left the settled count alone went with it: that call pokes the worker, so the worker may well make a pass before the next line runs. What shows it did not join the wait is the waiter count, which it leaves where it found it.

claude/workflow-reflex-integration-30ff2c

10 hours ago

bench: add repeated stateful page baseline for memo compilation (#7354) * bench: add repeated stateful page baseline for memo compilation * ci: expose fatal benchmark process diagnostics * ci: print native simulation logs after benchmark failures * ci: retain complete native benchmark crash diagnostics * ci: isolate compilation benchmarks from state profiling * ci: tolerate absent native benchmark logs * test: give process-tree sampling explicit CPU work * ci: remove temporary benchmark crash diagnostics

main

10 hours ago

Document session primitive regression tests

codex/codex/session-token-primitives

10 hours ago

Keep request cookies out of frontend state Preserve server-side cookie access while omitting Cookie from serialized router headers and raw_headers. Deprecate frontend router cookie reads in favor of rx.Cookie. Tests: 411 related unit tests passed; regression tests reproduced the leak before the fix. Ruff, formatting, and targeted pyright passed. Refs: ENG-12909

codex/session-cookie-privacy

10 hours ago

reflex-workflow: read the waiter cap rather than racing a crowd up to it The test started twelve callers and asserted four of them had come back inside two hundred milliseconds, which on a CI runner is a claim about how quickly asyncio schedules coroutines rather than about the cap. It failed on 3.10 the first time the postgres step ran, with none of the twelve finished. Set the count and call once. What is worth pinning is what a caller past the cap is told -- false, having watched nothing, and the worker asked to look anyway -- and that the same call under the cap waits and gets a pass, so the false is the cap talking. Both mutations, claiming success and dropping the cap, fail it.

claude/workflow-reflex-integration-30ff2c

10 hours ago

reflex-bench: expect the playground's fifth hydration frame in the wire app test on_load_internal now also updates the always dirty rx.dynamic var of /widgets, in a frame of its own before the delta setting is_hydrated. Claude-Session: https://claude.ai/code/session_01Y9bdwnG6ar7T4ay9nWmKb5

farhan/funny-ptolemy-2t8klw

10 hours ago

Set the macro job's timeout and budget from the warm run The warm run of the 7-benchmark macro suite billed 7:24 on the Cortex-A72 runner: 296 min a month for 30 daily and 10 label runs.

farhan/affectionate-dijkstra-dg3cgo

10 hours ago

ci: remove temporary benchmark crash diagnostics

codex/benchmark-repeated-stateful-page

10 hours ago

Latest Branches

0%

reflex-workflow: durable workflows on the application's own tables#7288

10 hours ago

50714c3

claude/workflow-reflex-integration-30ff2c

0%

Add signed browser sessions and bound client tokens#7361

10 hours ago

514269c

codex/codex/session-token-primitives

0%

Keep request cookies out of frontend state#7360

11 hours ago

27416d7

codex/session-cookie-privacy

© 2026 CodSpeed Technology

Home Terms Privacy Docs