Skip to content

build(deps): @effect/tsgo 0.63.0 (effect stays at 4.0.5) - #13

Merged
systemfsoftware-bot merged 4 commits into
mainfrom
deps/effect-4.1
Oct 29, 2026
Merged

systemfsoftware-bot merged 4 commits into
mainfrom
deps/effect-4.1

Conversation

@systemfsoftware-bot

@systemfsoftware-bot systemfsoftware-bot commented Oct 29, 2026 •

Copy link
Copy Markdown
Contributor

Upgrades @effect/tsgo to 0.63.0. effect stays at 4.0.5; every newer release breaks something here:

  • 4.1.0: alchemy (2.0.0-beta.80, the site's deploy tool) imports the effect/Crypto module that 4.1.0 removed.

  • 4.0.6: the storybook-gherkin failure corpus test loses the record's first location file (passes on 4.0.5, fails on 4.0.6).

  • 4.0.7: the effect-atom scenario 'Two requests for the next batch at the same time' deadlocks (passes on 4.0.6, fails on 4.0.7; Atom.js identical in both).

  • oxlint and @oxlint/plugins move to ~1.90.0 and oxlint-tsgolint to 7.0.2005, the range tsgo 0.63 supports.

  • pnpm overrides map effect, @effect/tsgo and the @effect/* packages the bundled tarballs and alchemy pull in to the catalog (4.0.5), so one version of each installs.

  • Schema.Class, TaggedClass and TaggedError get the brand type argument (schemaClassMissingBrand).

  • DuplicateCapture.name becomes capture and OutlineDuplicateRowName.name becomes rowName (errorPayloadShadowsName); changeset included.

  • Effect.flatMap with a re-yield becomes Effect.tap (flatMapToTap).

The Preview workflow fails on this and every PR because the repository has no Cloudflare secrets.

https://claude.ai/code/session_017rX35xUAHEjVVaDKVVzubW

Effect 4.1.0 is blocked: alchemy (through 2.0.0-beta.80) imports the removed effect/Crypto module.
Move oxlint to ~1.90.0 and oxlint-tsgolint to 7.0.2005 as tsgo 0.63 requires, map the
bundled tarballs' effect and @effect/* pins to the catalog, brand Schema classes,
rename shadowing error payload fields, and use Effect.tap where tsgo asks.

Claude-Session: https://claude.ai/code/session_017rX35xUAHEjVVaDKVVzubW
….0.5

4.0.6 and later read process.platform when the module loads. The vitest tarball imports
them in the browser project, where process is undefined, so every browser test failed to import.

Claude-Session: https://claude.ai/code/session_017rX35xUAHEjVVaDKVVzubW
effect 4.0.7 deadlocks the effect-atom scenario where two pulls of a feed wait on one signal
(it passes on 4.0.6 and fails on 4.0.7 with the same atom code).

Claude-Session: https://claude.ai/code/session_017rX35xUAHEjVVaDKVVzubW
@systemfsoftware-bot systemfsoftware-bot changed the title build(deps): effect 4.0.7 and @effect/tsgo 0.63.0 build(deps): effect 4.0.6 and @effect/tsgo 0.63.0 Oct 29, 2026
effect 4.0.6 breaks the storybook-gherkin failure corpus (the record loses its first location file)
and 4.0.7 deadlocks an effect-atom pull scenario; 4.1.0 is blocked by alchemy importing effect/Crypto.

Claude-Session: https://claude.ai/code/session_017rX35xUAHEjVVaDKVVzubW
@systemfsoftware-bot systemfsoftware-bot changed the title build(deps): effect 4.0.6 and @effect/tsgo 0.63.0 build(deps): @effect/tsgo 0.63.0 (effect stays at 4.0.5) Oct 29, 2026
@systemfsoftware-bot
systemfsoftware-bot merged commit ef4ec0e into main Oct 29, 2026
12 of 13 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant