fix: Remove 'use gpu' after transformation

[aleksanderkatan]

No description provided.

[github-actions]

pkg.pr.new

packages
Ready to be installed by your favorite package manager ⬇️

https://pkg.pr.new/software-mansion/TypeGPU/tgpu-gen@985409ae612431f0a0c16e8275ef3eee07118b86

https://pkg.pr.new/software-mansion/TypeGPU/tinyest-for-wgsl@985409ae612431f0a0c16e8275ef3eee07118b86

https://pkg.pr.new/software-mansion/TypeGPU/typegpu@985409ae612431f0a0c16e8275ef3eee07118b86

https://pkg.pr.new/software-mansion/TypeGPU/@typegpu/cli@985409ae612431f0a0c16e8275ef3eee07118b86

https://pkg.pr.new/software-mansion/TypeGPU/@typegpu/color@985409ae612431f0a0c16e8275ef3eee07118b86

https://pkg.pr.new/software-mansion/TypeGPU/@typegpu/gl@985409ae612431f0a0c16e8275ef3eee07118b86

https://pkg.pr.new/software-mansion/TypeGPU/@typegpu/noise@985409ae612431f0a0c16e8275ef3eee07118b86

https://pkg.pr.new/software-mansion/TypeGPU/@typegpu/radiance-cascades@985409ae612431f0a0c16e8275ef3eee07118b86

https://pkg.pr.new/software-mansion/TypeGPU/@typegpu/react@985409ae612431f0a0c16e8275ef3eee07118b86

https://pkg.pr.new/software-mansion/TypeGPU/@typegpu/sdf@985409ae612431f0a0c16e8275ef3eee07118b86

https://pkg.pr.new/software-mansion/TypeGPU/@typegpu/three@985409ae612431f0a0c16e8275ef3eee07118b86

https://pkg.pr.new/software-mansion/TypeGPU/unplugin-typegpu@985409ae612431f0a0c16e8275ef3eee07118b86

benchmark
view benchmark

commit
view commit

[github-actions]

Bundle size comparison (import * as ... in PR vs import * as ... in target):

🟢 Decreased (max -6.74%)	➖ Unchanged	🔴 Increased	❔ Unknown
1	320	0	0

import * as ... in PR vs import * as ... in target (did bundle size increase?):

Test	tsdown
STATIC_useGpuFunction.ts	263 B ($${\color{green}-6.7\%}$$)

import { ... } in PR vs import * as ... in PR (is the library tree-Shakeable?):

Test	tsdown
tgpu_init.ts	256.45 kB ($${\color{green}-3.2\%}$$)
tgpu_initFromDevice.ts	255.93 kB ($${\color{green}-3.4\%}$$)
tgpu_resolve.ts	187.16 kB ($${\color{green}-29.3\%}$$)
tgpu_resolveWithContext.ts	187.10 kB ($${\color{green}-29.4\%}$$)
tgpu_bindGroupLayout.ts	69.01 kB ($${\color{green}-74.0\%}$$)
tgpu_mutableAccessor.ts	65.98 kB ($${\color{green}-75.1\%}$$)
tgpu_accessor.ts	65.97 kB ($${\color{green}-75.1\%}$$)
tgpu_privateVar.ts	65.30 kB ($${\color{green}-75.3\%}$$)
tgpu_workgroupVar.ts	65.30 kB ($${\color{green}-75.4\%}$$)
tgpu_const.ts	64.53 kB ($${\color{green}-75.6\%}$$)
tgpu_fn.ts	37.97 kB ($${\color{green}-85.7\%}$$)
tgpu_fragmentFn.ts	37.91 kB ($${\color{green}-85.7\%}$$)
tgpu_vertexFn.ts	37.72 kB ($${\color{green}-85.8\%}$$)
tgpu_computeFn.ts	37.43 kB ($${\color{green}-85.9\%}$$)
tgpu_vertexLayout.ts	26.61 kB ($${\color{green}-90.0\%}$$)
tgpu_comptime.ts	14.41 kB ($${\color{green}-94.6\%}$$)
tgpu_unroll.ts	2.60 kB ($${\color{green}-99.0\%}$$)
tgpu_slot.ts	1.57 kB ($${\color{green}-99.4\%}$$)
tgpu_lazy.ts	1.22 kB ($${\color{green}-99.5\%}$$)

If you wish to run a comparison for other, slower bundlers, run the 'Tree-shake test' from the GitHub Actions menu.

[github-actions]

Resolution Time Benchmark

---
config:
  themeVariables:
    xyChart:
      plotColorPalette: "#E63946, #3B82F6, #059669"
---
xychart
  title "Random Branching (🔴 PR | 🔵 main | 🟢 release)"
  x-axis "max depth" [1, 2, 3, 4, 5, 6, 7, 8]
  y-axis "time (ms)"
  line [0.80, 1.65, 3.31, 5.29, 6.68, 10.82, 18.68, 19.70]
  line [0.87, 1.69, 3.85, 5.64, 6.42, 10.23, 19.79, 21.99]
  line [0.85, 1.64, 3.61, 5.73, 7.27, 10.58, 18.91, 19.57]

Loading

---
config:
  themeVariables:
    xyChart:
      plotColorPalette: "#E63946, #3B82F6, #059669"
---
xychart
  title "Linear Recursion (🔴 PR | 🔵 main | 🟢 release)"
  x-axis "max depth" [1, 2, 3, 4, 5, 6, 7, 8]
  y-axis "time (ms)"
  line [0.27, 0.50, 0.60, 0.70, 0.98, 0.96, 1.27, 1.37]
  line [0.29, 0.49, 0.71, 0.76, 1.04, 1.09, 1.27, 1.42]
  line [0.28, 0.50, 0.61, 0.75, 1.00, 1.06, 1.27, 1.41]

Loading

---
config:
  themeVariables:
    xyChart:
      plotColorPalette: "#E63946, #3B82F6, #059669"
---
xychart
  title "Full Tree (🔴 PR | 🔵 main | 🟢 release)"
  x-axis "max depth" [1, 2, 3, 4, 5, 6, 7, 8]
  y-axis "time (ms)"
  line [0.80, 1.90, 4.33, 6.16, 11.75, 24.06, 48.56, 100.40]
  line [0.87, 1.88, 4.03, 5.50, 10.91, 23.79, 48.61, 99.19]
  line [0.84, 1.88, 3.80, 5.72, 11.05, 23.25, 49.95, 100.05]

Loading

[aleksanderkatan]

@pullfrog review

[pullfrog]

✅ No new issues found.

Reviewed changes — This PR removes the custom "use gpu" directive from functions after TypeGPU transforms them, replacing the previous alreadyTransformed WeakSet guard with directive deletion to prevent re-transformation.

• Remove "use gpu" after transformation in both the Babel and unplugin/rollup plugin paths by adding a removeUseGpuDirective method to TransformMethods.
• Drop the alreadyTransformed state from PluginState; since the directive is stripped, a second pass no longer recognizes the function as a GPU target.
• Add test coverage for directive removal, coexistence with other directives like 'worklet', and double transformation round-trips for both backends.
• Update inline snapshots across auto-naming, typescript-syntax, and autoname tests to reflect the absence of the leftover directive.

  ^{｜ View workflow run ｜ Using Kimi K2 (free via Pullfrog for OSS) ｜ 𝕏}

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.

[pullfrog]

✅ No new issues found.

Reviewed changes — this PR removes the 'use gpu' directive from transformed functions so a second transform pass doesn't re-process them, replacing the previous alreadyTransformed WeakSet guard.

• Updated the Babel transform (packages/unplugin-typegpu/src/babel.ts) by adding removeUseGpuDirective and dropping the alreadyTransformed check in assignMetadata.
• Updated the unplugin/Rollup transform (packages/unplugin-typegpu/src/core/factory.ts) with a MagicString-based directive removal implementation.
• Updated shared core state (packages/unplugin-typegpu/src/core/common.ts) to expose removeUseGpuDirective and remove the now-unused alreadyTransformed state.
• Refreshed inline snapshots in auto-naming, TypeScript-syntax, and the internal TypeGPU autoname test.
• Added regression tests covering directive removal (with siblings preserved) and double transformation for both Babel and Rollup.

  ^{｜ View workflow run ｜ Using Kimi K2 (free via Pullfrog for OSS) ｜ 𝕏}