client-core: gate the host-process modules behind native #205

Merged

natsukium merged 2 commits from fix/portable-core-build-195 into main 2026-09-07 22:52:35 +09:00

Conversation 0 Commits 2 Files changed 6 +26 -12

natsukium commented 2026-09-07 22:29:49 +09:00

Owner

Copy link

Fixes #195.

env_base, shader_clock, and pipe were declared unconditionally while
reaching for native-only items (connector::Carrier, config::ShaderAnimation)
and cfg(unix|windows) helpers, so felis-client-core with
default-features = false — the documented reuse path for a browser client —
failed to compile on wasm32 and on the host alike. All three are host-process
concerns and their only consumers build with default features, so they are now
gated behind native.

The regression survived three weeks because no gate ever built that
configuration: every CI job runs --all-features. pr.yml and the new
just check-portable recipe (wired into just check) now lint
-p felis-client-core --no-default-features --target wasm32-unknown-unknown.
Clippy rather than check, so -D warnings also catches the dead cfg
leftovers a feature split strands.

Doc cascade: docs/reference/testing.md records the new gate in the check
tables, and CONTRIBUTING.md follows the same list.

Verified: just check green (fmt, clippy, nextest, deny, proto-compat); reviewed
by pi luna + pi sol; docs proofread by Gemini gemini-3.8-flash-high

Deferred

• Extend the wasm32 guard to felis-grid and felis-protocol, the other crates the
  portable core pulls in — they compile clean for wasm32 today as a transitive
  consequence of the felis-client-core guard, so a separate step would add CI
  time without adding coverage.
• Document the portable module set in docs/reference/workspace.md's reuse table
  — the table names the crates reused per client kind, not their modules; the
  module-level invariant now lives in the crate doc comment, which is where a
  contributor editing lib.rs will see it.

Fixes #195. `env_base`, `shader_clock`, and `pipe` were declared unconditionally while reaching for `native`-only items (`connector::Carrier`, `config::ShaderAnimation`) and `cfg(unix|windows)` helpers, so `felis-client-core` with `default-features = false` — the documented reuse path for a browser client — failed to compile on wasm32 and on the host alike. All three are host-process concerns and their only consumers build with default features, so they are now gated behind `native`. The regression survived three weeks because no gate ever built that configuration: every CI job runs `--all-features`. `pr.yml` and the new `just check-portable` recipe (wired into `just check`) now lint `-p felis-client-core --no-default-features --target wasm32-unknown-unknown`. Clippy rather than `check`, so `-D warnings` also catches the dead `cfg` leftovers a feature split strands. Doc cascade: `docs/reference/testing.md` records the new gate in the check tables, and `CONTRIBUTING.md` follows the same list. Verified: just check green (fmt, clippy, nextest, deny, proto-compat); reviewed by pi luna + pi sol; docs proofread by Gemini gemini-3.8-flash-high ## Deferred - Extend the wasm32 guard to felis-grid and felis-protocol, the other crates the portable core pulls in — they compile clean for wasm32 today as a transitive consequence of the felis-client-core guard, so a separate step would add CI time without adding coverage. - Document the portable module set in `docs/reference/workspace.md`'s reuse table — the table names the crates reused per client kind, not their modules; the module-level invariant now lives in the crate doc comment, which is where a contributor editing `lib.rs` will see it.

natsukium added 2 commits 2026-09-07 22:29:49 +09:00

client-core: gate the host-process modules behind native ... 60df822da9

`env_base`, `shader_clock`, and `pipe` were declared unconditionally
while reaching for `native`-only items (`connector::Carrier`,
`config::ShaderAnimation`) and `cfg(unix|windows)`-only helpers, so
`default-features = false` failed to compile on wasm32 and on the host
alike. That build is the documented reuse path for a browser client, and
felis-web-component cannot build against felis at all while it is broken.

All three are host-process concerns; their only consumers (felis-cli,
felis-client, and this crate's own `dial`) build with default features.
Splitting `pipe` into a portable half was rejected: `Origin` and
`StagedRegion` describe a spawn on this machine, and no portable
consumer exists for them.

Refs #195

ci: guard felis-client-core's portable core on wasm32 ... d1b9510a16

The wasm32 target has been on the toolchain since the feature split, with
`dev/flake-module.nix` claiming CI guards the portable core against
bitrot, but nothing ever built that configuration: every gate runs
`--all-features`. Three separate regressions landed in three weeks
before a downstream reported the crate no longer compiled.

Clippy rather than check, so `-D warnings` also catches the dead `cfg`
leftovers a feature split strands, which `cargo check` accepts silently.

Refs #195

natsukium merged commit d1b9510a16 into main 2026-09-07 22:52:35 +09:00

natsukium deleted branch fix/portable-core-build-195 2026-09-07 22:52:35 +09:00

Sign in to join this conversation.

Reviewers

No reviewers

Labels

Clear labels   

priority/P0

Must be resolved before the v0.1.0 tag

  

priority/P1

Must be settled before the first public contract freezes

  

priority/P2

Explicitly deferred until after v0.1.0

  

release/v0.1.0

Work tracked for the first tagged release

  

status/blocked

Cannot start now: waits on an upstream release, another issue, or a recorded revisit trigger

  

status/planned

Triaged with a source-grounded plan and a passed pi review; ready to implement

  

type/bug

Shipped behavior that contradicts the docs or the code's own intent; needs a product code change

  

type/design

An open scope or trade-off question; needs a recorded decision in the owning explanation doc before code

  

type/test-gap

Documented or reviewed behavior that no test pins; usually split from a review finding, needs no product code change

  

type/tracker

Parent issue that tracks child work

No labels

priority/P0

priority/P1

priority/P2

release/v0.1.0

status/blocked

status/planned

type/bug

type/design

type/test-gap

type/tracker

Milestone

Clear milestone

No items

No milestone

Projects

Clear projects

No items

No project

Assignees

Clear assignees

No assignees

1 participant

Notifications

Subscribe

Due date

The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set

Reference

natsukium/felis!205

No description provided.