fix: improve runtime ternary support by refining side-effect detection

[pullfrog]

Summary

Fixes runtime ternary operator support by correcting side-effect detection across multiple code paths. Addresses all failing cases from #2587.

Changes

Bug fixes (source)

• resolutionCtx.ts: Function arguments (snip(name, type, origin)) now correctly set possibleSideEffects: false — reading a function parameter is always side-effect-free
• tgpuConstant.ts: tgpu.const().$ proxy snippet now correctly set possibleSideEffects: false — reading a compile-time constant is side-effect-free
• wgslGenerator.ts: Pure binary operators (comparisons, bit shifts, bitwise, modulo) now propagate operand side-effects instead of always defaulting to true
• dualImpl.ts: Added missing side-effect tracking — dualImpl callable functions now return noSideEffects(result) when all arguments are side-effect-free, matching the behavior already present in callableSchema
• bufferUsage.ts: Buffer read accesses (uniform, readonly, mutable) now correctly set possibleSideEffects: false

Tests

• ternaryRuntime.test.ts (new): 8 tests covering all failing cases from the issue, including:
  • Subtraction in branches with function parameters
  • tgpu.const array indexing in branches
  • Nested runtime ternaries
  • Bit shift in branches
  • Struct field access across ternaries
  • Buffer layout access in ternary branches
  • Ternary with assignment in branch (should throw)
  • Ternary with comparison and unary negation
• ternaryOperator.test.ts: Updated the "should throw" test to "should generate select()" since n > 0 ? n : -n now correctly produces a select() call

Verification

• All 2018 tests pass (155 test files)
• TypeScript typecheck passes (tsc --noEmit)

Closes #2587

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

[Copilot]

Copilot encountered an error and was unable to review this pull request. You can try again by re-requesting a review.

[github-actions]

pkg.pr.new

packages
Ready to be installed by your favorite package manager ⬇️

https://pkg.pr.new/software-mansion/TypeGPU/typegpu@c7de770c8bb4f49f6a108a7ec0691f6901b78787

https://pkg.pr.new/software-mansion/TypeGPU/@typegpu/cli@c7de770c8bb4f49f6a108a7ec0691f6901b78787

https://pkg.pr.new/software-mansion/TypeGPU/@typegpu/noise@c7de770c8bb4f49f6a108a7ec0691f6901b78787

https://pkg.pr.new/software-mansion/TypeGPU/@typegpu/react@c7de770c8bb4f49f6a108a7ec0691f6901b78787

https://pkg.pr.new/software-mansion/TypeGPU/unplugin-typegpu@c7de770c8bb4f49f6a108a7ec0691f6901b78787

benchmark
view benchmark

commit
view commit

[github-actions]

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

🟢 Decreased	➖ Unchanged	🔴 Increased	❔ Unknown
0	135	185	0

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

Click to reveal the results table (163 entries).

Test	tsdown
STATIC_std.ts	108.25 kB ($${\color{red}+2.2\%}$$)
std_dot.ts	67.14 kB ($${\color{red}+1.7\%}$$)
std_length.ts	67.14 kB ($${\color{red}+1.7\%}$$)
std_distance.ts	67.14 kB ($${\color{red}+1.7\%}$$)
std_dot4U8Packed.ts	67.14 kB ($${\color{red}+1.7\%}$$)
std_fract.ts	67.14 kB ($${\color{red}+1.7\%}$$)
std_frexp.ts	67.14 kB ($${\color{red}+1.7\%}$$)
std_mix.ts	67.14 kB ($${\color{red}+1.7\%}$$)
std_modf.ts	67.14 kB ($${\color{red}+1.7\%}$$)
std_abs.ts	67.15 kB ($${\color{red}+1.7\%}$$)
std_acos.ts	67.15 kB ($${\color{red}+1.7\%}$$)
std_acosh.ts	67.15 kB ($${\color{red}+1.7\%}$$)
std_asin.ts	67.15 kB ($${\color{red}+1.7\%}$$)
std_asinh.ts	67.15 kB ($${\color{red}+1.7\%}$$)
std_atan.ts	67.15 kB ($${\color{red}+1.7\%}$$)
std_atan2.ts	67.15 kB ($${\color{red}+1.7\%}$$)
std_atanh.ts	67.15 kB ($${\color{red}+1.7\%}$$)
std_ceil.ts	67.15 kB ($${\color{red}+1.7\%}$$)
std_clamp.ts	67.15 kB ($${\color{red}+1.7\%}$$)
std_cos.ts	67.15 kB ($${\color{red}+1.7\%}$$)
std_cosh.ts	67.15 kB ($${\color{red}+1.7\%}$$)
std_countLeadingZeros.ts	67.15 kB ($${\color{red}+1.7\%}$$)
std_countOneBits.ts	67.15 kB ($${\color{red}+1.7\%}$$)
std_countTrailingZeros.ts	67.15 kB ($${\color{red}+1.7\%}$$)
std_cross.ts	67.15 kB ($${\color{red}+1.7\%}$$)
std_degrees.ts	67.15 kB ($${\color{red}+1.7\%}$$)
std_determinant.ts	67.15 kB ($${\color{red}+1.7\%}$$)
std_dot4I8Packed.ts	67.15 kB ($${\color{red}+1.7\%}$$)
std_exp.ts	67.15 kB ($${\color{red}+1.7\%}$$)
std_exp2.ts	67.15 kB ($${\color{red}+1.7\%}$$)
std_extractBits.ts	67.15 kB ($${\color{red}+1.7\%}$$)
std_faceForward.ts	67.15 kB ($${\color{red}+1.7\%}$$)
std_firstLeadingBit.ts	67.15 kB ($${\color{red}+1.7\%}$$)
std_firstTrailingBit.ts	67.15 kB ($${\color{red}+1.7\%}$$)
std_floor.ts	67.15 kB ($${\color{red}+1.7\%}$$)
std_fma.ts	67.15 kB ($${\color{red}+1.7\%}$$)
std_insertBits.ts	67.15 kB ($${\color{red}+1.7\%}$$)
std_inverseSqrt.ts	67.15 kB ($${\color{red}+1.7\%}$$)
std_ldexp.ts	67.15 kB ($${\color{red}+1.7\%}$$)
std_log.ts	67.15 kB ($${\color{red}+1.7\%}$$)
std_log2.ts	67.15 kB ($${\color{red}+1.7\%}$$)
std_max.ts	67.15 kB ($${\color{red}+1.7\%}$$)
std_min.ts	67.15 kB ($${\color{red}+1.7\%}$$)
std_normalize.ts	67.15 kB ($${\color{red}+1.7\%}$$)
std_pow.ts	67.15 kB ($${\color{red}+1.7\%}$$)
std_quantizeToF16.ts	67.15 kB ($${\color{red}+1.7\%}$$)
std_radians.ts	67.15 kB ($${\color{red}+1.7\%}$$)
std_reflect.ts	67.15 kB ($${\color{red}+1.7\%}$$)
std_refract.ts	67.15 kB ($${\color{red}+1.7\%}$$)
std_reverseBits.ts	67.15 kB ($${\color{red}+1.7\%}$$)
std_round.ts	67.15 kB ($${\color{red}+1.7\%}$$)
std_saturate.ts	67.15 kB ($${\color{red}+1.7\%}$$)
std_sign.ts	67.15 kB ($${\color{red}+1.7\%}$$)
std_sin.ts	67.15 kB ($${\color{red}+1.7\%}$$)
std_sinh.ts	67.15 kB ($${\color{red}+1.7\%}$$)
std_smoothstep.ts	67.15 kB ($${\color{red}+1.7\%}$$)
std_sqrt.ts	67.15 kB ($${\color{red}+1.7\%}$$)
std_step.ts	67.15 kB ($${\color{red}+1.7\%}$$)
std_tan.ts	67.15 kB ($${\color{red}+1.7\%}$$)
std_tanh.ts	67.15 kB ($${\color{red}+1.7\%}$$)
std_transpose.ts	67.15 kB ($${\color{red}+1.7\%}$$)
std_trunc.ts	67.15 kB ($${\color{red}+1.7\%}$$)
std_subgroupAdd.ts	24.34 kB ($${\color{red}+1.5\%}$$)
std_subgroupAll.ts	24.35 kB ($${\color{red}+1.5\%}$$)
std_subgroupAnd.ts	24.35 kB ($${\color{red}+1.5\%}$$)
std_subgroupAny.ts	24.35 kB ($${\color{red}+1.5\%}$$)
std_subgroupBallot.ts	24.35 kB ($${\color{red}+1.5\%}$$)
std_subgroupBroadcast.ts	24.35 kB ($${\color{red}+1.5\%}$$)
std_subgroupBroadcastFirst.ts	24.35 kB ($${\color{red}+1.5\%}$$)
std_subgroupElect.ts	24.35 kB ($${\color{red}+1.5\%}$$)
std_subgroupExclusiveAdd.ts	24.35 kB ($${\color{red}+1.5\%}$$)
std_subgroupExclusiveMul.ts	24.35 kB ($${\color{red}+1.5\%}$$)
std_subgroupInclusiveAdd.ts	24.35 kB ($${\color{red}+1.5\%}$$)
std_subgroupInclusiveMul.ts	24.35 kB ($${\color{red}+1.5\%}$$)
std_subgroupMax.ts	24.35 kB ($${\color{red}+1.5\%}$$)
std_subgroupMin.ts	24.35 kB ($${\color{red}+1.5\%}$$)
std_subgroupMul.ts	24.35 kB ($${\color{red}+1.5\%}$$)
std_subgroupOr.ts	24.35 kB ($${\color{red}+1.5\%}$$)
std_subgroupShuffle.ts	24.35 kB ($${\color{red}+1.5\%}$$)
std_subgroupShuffleDown.ts	24.35 kB ($${\color{red}+1.5\%}$$)
std_subgroupShuffleUp.ts	24.35 kB ($${\color{red}+1.5\%}$$)
std_subgroupShuffleXor.ts	24.35 kB ($${\color{red}+1.5\%}$$)
std_subgroupXor.ts	24.35 kB ($${\color{red}+1.5\%}$$)
std_atomicLoad.ts	16.02 kB ($${\color{red}+1.5\%}$$)
std_atomicStore.ts	16.02 kB ($${\color{red}+1.5\%}$$)
std_textureBarrier.ts	16.02 kB ($${\color{red}+1.5\%}$$)
std_atomicAdd.ts	16.03 kB ($${\color{red}+1.5\%}$$)
std_atomicAnd.ts	16.03 kB ($${\color{red}+1.5\%}$$)
std_atomicMax.ts	16.03 kB ($${\color{red}+1.5\%}$$)
std_atomicMin.ts	16.03 kB ($${\color{red}+1.5\%}$$)
std_atomicOr.ts	16.03 kB ($${\color{red}+1.5\%}$$)
std_atomicSub.ts	16.03 kB ($${\color{red}+1.5\%}$$)
std_atomicXor.ts	16.03 kB ($${\color{red}+1.5\%}$$)
std_storageBarrier.ts	16.03 kB ($${\color{red}+1.5\%}$$)
std_workgroupBarrier.ts	16.03 kB ($${\color{red}+1.5\%}$$)
std_dpdx.ts	15.22 kB ($${\color{red}+1.2\%}$$)
std_dpdxCoarse.ts	15.22 kB ($${\color{red}+1.2\%}$$)
std_dpdxFine.ts	15.22 kB ($${\color{red}+1.2\%}$$)
std_dpdy.ts	15.22 kB ($${\color{red}+1.2\%}$$)
std_dpdyCoarse.ts	15.22 kB ($${\color{red}+1.2\%}$$)
std_dpdyFine.ts	15.22 kB ($${\color{red}+1.2\%}$$)
std_fwidth.ts	15.22 kB ($${\color{red}+1.2\%}$$)
std_fwidthCoarse.ts	15.22 kB ($${\color{red}+1.2\%}$$)
std_fwidthFine.ts	15.22 kB ($${\color{red}+1.2\%}$$)
std_and.ts	52.19 kB ($${\color{red}+0.9\%}$$)
std_or.ts	52.19 kB ($${\color{red}+0.9\%}$$)
std_all.ts	52.19 kB ($${\color{red}+0.9\%}$$)
std_allEq.ts	52.19 kB ($${\color{red}+0.9\%}$$)
std_eq.ts	52.19 kB ($${\color{red}+0.9\%}$$)
std_lt.ts	52.19 kB ($${\color{red}+0.9\%}$$)
std_ne.ts	52.19 kB ($${\color{red}+0.9\%}$$)
std_not.ts	52.19 kB ($${\color{red}+0.9\%}$$)
std_select.ts	52.19 kB ($${\color{red}+0.9\%}$$)
std_any.ts	52.20 kB ($${\color{red}+0.9\%}$$)
std_ge.ts	52.20 kB ($${\color{red}+0.9\%}$$)
std_gt.ts	52.20 kB ($${\color{red}+0.9\%}$$)
std_isCloseTo.ts	52.20 kB ($${\color{red}+0.9\%}$$)
std_le.ts	52.20 kB ($${\color{red}+0.9\%}$$)
std_textureDimensions.ts	25.96 kB ($${\color{red}+0.8\%}$$)
std_textureGather.ts	25.96 kB ($${\color{red}+0.8\%}$$)
std_textureLoad.ts	25.96 kB ($${\color{red}+0.8\%}$$)
std_textureSample.ts	25.96 kB ($${\color{red}+0.8\%}$$)
std_textureSampleBaseClampToEdge.ts	25.96 kB ($${\color{red}+0.8\%}$$)
std_textureSampleBias.ts	25.96 kB ($${\color{red}+0.8\%}$$)
std_textureSampleCompare.ts	25.96 kB ($${\color{red}+0.8\%}$$)
std_textureSampleCompareLevel.ts	25.96 kB ($${\color{red}+0.8\%}$$)
std_textureSampleGrad.ts	25.96 kB ($${\color{red}+0.8\%}$$)
std_textureSampleLevel.ts	25.96 kB ($${\color{red}+0.8\%}$$)
std_textureStore.ts	25.96 kB ($${\color{red}+0.8\%}$$)
STATIC_allImports.ts	293.83 kB ($${\color{red}+0.8\%}$$)
std_rotateX4.ts	49.44 kB ($${\color{red}+0.6\%}$$)
std_scale4.ts	49.44 kB ($${\color{red}+0.6\%}$$)
std_translate4.ts	49.44 kB ($${\color{red}+0.6\%}$$)
std_rotateY4.ts	49.45 kB ($${\color{red}+0.6\%}$$)
std_rotateZ4.ts	49.45 kB ($${\color{red}+0.6\%}$$)
STATIC_tgpu.ts	263.08 kB ($${\color{red}+0.6\%}$$)
tgpu_fn.ts	263.08 kB ($${\color{red}+0.6\%}$$)
tgpu_init.ts	263.09 kB ($${\color{red}+0.6\%}$$)
tgpu_lazy.ts	263.09 kB ($${\color{red}+0.6\%}$$)
tgpu_slot.ts	263.09 kB ($${\color{red}+0.6\%}$$)
tgpu_const.ts	263.09 kB ($${\color{red}+0.6\%}$$)
tgpu_unroll.ts	263.09 kB ($${\color{red}+0.6\%}$$)
tgpu_resolve.ts	263.09 kB ($${\color{red}+0.6\%}$$)
tgpu_accessor.ts	263.09 kB ($${\color{red}+0.6\%}$$)
tgpu_comptime.ts	263.09 kB ($${\color{red}+0.6\%}$$)
tgpu_vertexFn.ts	263.09 kB ($${\color{red}+0.6\%}$$)
tgpu_computeFn.ts	263.09 kB ($${\color{red}+0.6\%}$$)
tgpu_fragmentFn.ts	263.09 kB ($${\color{red}+0.6\%}$$)
tgpu_privateVar.ts	263.09 kB ($${\color{red}+0.6\%}$$)
tgpu_vertexLayout.ts	263.09 kB ($${\color{red}+0.6\%}$$)
tgpu_workgroupVar.ts	263.09 kB ($${\color{red}+0.6\%}$$)
tgpu_initFromDevice.ts	263.10 kB ($${\color{red}+0.6\%}$$)
tgpu_bindGroupLayout.ts	263.10 kB ($${\color{red}+0.6\%}$$)
tgpu_mutableAccessor.ts	263.10 kB ($${\color{red}+0.6\%}$$)
tgpu_resolveWithContext.ts	263.10 kB ($${\color{red}+0.6\%}$$)
std_neg.ts	48.60 kB ($${\color{red}+0.5\%}$$)
std_add.ts	48.61 kB ($${\color{red}+0.5\%}$$)
std_bitShiftLeft.ts	48.61 kB ($${\color{red}+0.5\%}$$)
std_bitShiftRight.ts	48.61 kB ($${\color{red}+0.5\%}$$)
std_div.ts	48.61 kB ($${\color{red}+0.5\%}$$)
std_mod.ts	48.61 kB ($${\color{red}+0.5\%}$$)
std_mul.ts	48.61 kB ($${\color{red}+0.5\%}$$)
std_sub.ts	48.61 kB ($${\color{red}+0.5\%}$$)

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

Click to reveal the results table (297 entries).

Test	tsdown
d_Void.ts	87.61 kB ($${\color{red}+12121.8\%}$$)
d_sampler.ts	87.61 kB ($${\color{red}+11990.4\%}$$)
d_isPtr.ts	87.61 kB ($${\color{red}+11829.4\%}$$)
d_comparisonSampler.ts	87.62 kB ($${\color{red}+11815.1\%}$$)
d_isSizeAttrib.ts	87.61 kB ($${\color{red}+11798.7\%}$$)
d_isWgslArray.ts	87.61 kB ($${\color{red}+11798.5\%}$$)
d_isAlignAttrib.ts	87.61 kB ($${\color{red}+11783.0\%}$$)
d_isWgslStruct.ts	87.61 kB ($${\color{red}+11782.9\%}$$)
d_isAtomic.ts	87.61 kB ($${\color{red}+11782.4\%}$$)
d_isBuiltinAttrib.ts	87.62 kB ($${\color{red}+11751.9\%}$$)
d_isLocationAttrib.ts	87.62 kB ($${\color{red}+11736.4\%}$$)
d_isDecorated.ts	87.61 kB ($${\color{red}+11735.8\%}$$)
d_isInterpolateAttrib.ts	87.62 kB ($${\color{red}+11690.1\%}$$)
d_atomic.ts	87.61 kB ($${\color{red}+11416.0\%}$$)
d_ptrHandle.ts	87.61 kB ($${\color{red}+10442.1\%}$$)
d_ptrUniform.ts	87.61 kB ($${\color{red}+10429.8\%}$$)
d_ptrStorage.ts	87.61 kB ($${\color{red}+10380.6\%}$$)
d_ptrPrivate.ts	87.61 kB ($${\color{red}+10356.2\%}$$)
d_ptrFn.ts	87.61 kB ($${\color{red}+10343.4\%}$$)
d_ptrWorkgroup.ts	87.61 kB ($${\color{red}+10332.1\%}$$)
d_textureExternal.ts	87.62 kB ($${\color{red}+10177.1\%}$$)
d_textureStorage1d.ts	87.62 kB ($${\color{red}+8551.9\%}$$)
d_textureStorage2d.ts	87.62 kB ($${\color{red}+8551.9\%}$$)
d_textureStorage3d.ts	87.62 kB ($${\color{red}+8551.9\%}$$)
d_textureStorage2dArray.ts	87.62 kB ($${\color{red}+8404.7\%}$$)
d_isDisarray.ts	87.61 kB ($${\color{red}+7776.6\%}$$)
d_isUnstruct.ts	87.61 kB ($${\color{red}+7776.6\%}$$)
d_isLooseDecorated.ts	87.62 kB ($${\color{red}+7729.0\%}$$)
d_isLooseData.ts	87.61 kB ($${\color{red}+7464.5\%}$$)
d_isWgslData.ts	87.61 kB ($${\color{red}+6585.1\%}$$)
d_unstruct.ts	87.61 kB ($${\color{red}+5560.1\%}$$)
d_isData.ts	87.61 kB ($${\color{red}+4746.6\%}$$)
d_deepEqual.ts	87.61 kB ($${\color{red}+3896.1\%}$$)
d_struct.ts	87.61 kB ($${\color{red}+2330.5\%}$$)
d_ref.ts	87.60 kB ($${\color{red}+1729.6\%}$$)
std_discard.ts	105.42 kB ($${\color{red}+639.4\%}$$)
std_isBeingTranspiled.ts	105.43 kB ($${\color{red}+628.8\%}$$)
std_copy.ts	105.42 kB ($${\color{red}+627.0\%}$$)
std_arrayLength.ts	105.42 kB ($${\color{red}+626.9\%}$$)
std_getTargetShaderLanguage.ts	105.43 kB ($${\color{red}+625.5\%}$$)
std_extensionEnabled.ts	105.43 kB ($${\color{red}+622.9\%}$$)
std_range.ts	105.42 kB ($${\color{red}+609.8\%}$$)
std_dpdx.ts	105.42 kB ($${\color{red}+592.7\%}$$)
std_fwidthCoarse.ts	105.42 kB ($${\color{red}+592.5\%}$$)
std_dpdxCoarse.ts	105.42 kB ($${\color{red}+592.5\%}$$)
std_dpdyCoarse.ts	105.42 kB ($${\color{red}+592.5\%}$$)
std_fwidthFine.ts	105.42 kB ($${\color{red}+592.5\%}$$)
std_dpdxFine.ts	105.42 kB ($${\color{red}+592.5\%}$$)
std_dpdyFine.ts	105.42 kB ($${\color{red}+592.5\%}$$)
std_fwidth.ts	105.42 kB ($${\color{red}+592.5\%}$$)
std_dpdy.ts	105.42 kB ($${\color{red}+592.4\%}$$)
d_bool.ts	87.61 kB ($${\color{red}+575.2\%}$$)
d_f16.ts	87.60 kB ($${\color{red}+575.2\%}$$)
d_f32.ts	87.60 kB ($${\color{red}+575.2\%}$$)
d_i32.ts	87.60 kB ($${\color{red}+575.2\%}$$)
d_u32.ts	87.60 kB ($${\color{red}+575.2\%}$$)
d_u16.ts	87.60 kB ($${\color{red}+574.0\%}$$)
std_atomicLoad.ts	105.42 kB ($${\color{red}+558.1\%}$$)
std_textureBarrier.ts	105.42 kB ($${\color{red}+557.9\%}$$)
std_atomicStore.ts	105.42 kB ($${\color{red}+557.9\%}$$)
std_workgroupBarrier.ts	105.43 kB ($${\color{red}+557.7\%}$$)
std_storageBarrier.ts	105.42 kB ($${\color{red}+557.7\%}$$)
std_atomicAdd.ts	105.42 kB ($${\color{red}+557.6\%}$$)
std_atomicAnd.ts	105.42 kB ($${\color{red}+557.6\%}$$)
std_atomicMax.ts	105.42 kB ($${\color{red}+557.6\%}$$)
std_atomicMin.ts	105.42 kB ($${\color{red}+557.6\%}$$)
std_atomicSub.ts	105.42 kB ($${\color{red}+557.6\%}$$)
std_atomicXor.ts	105.42 kB ($${\color{red}+557.6\%}$$)
std_atomicOr.ts	105.42 kB ($${\color{red}+557.6\%}$$)
d_textureDepth2d.ts	87.62 kB ($${\color{red}+553.0\%}$$)
d_textureDepthCube.ts	87.62 kB ($${\color{red}+552.7\%}$$)
d_texture1d.ts	87.61 kB ($${\color{red}+552.4\%}$$)
d_texture2d.ts	87.61 kB ($${\color{red}+552.4\%}$$)
d_texture3d.ts	87.61 kB ($${\color{red}+552.4\%}$$)
d_textureDepth2dArray.ts	87.62 kB ($${\color{red}+552.2\%}$$)
d_textureCube.ts	87.61 kB ($${\color{red}+552.1\%}$$)
d_textureDepthCubeArray.ts	87.62 kB ($${\color{red}+551.9\%}$$)
d_textureDepthMultisampled2d.ts	87.63 kB ($${\color{red}+551.8\%}$$)
d_texture2dArray.ts	87.62 kB ($${\color{red}+551.6\%}$$)
d_textureCubeArray.ts	87.62 kB ($${\color{red}+551.3\%}$$)
d_textureMultisampled2d.ts	87.62 kB ($${\color{red}+551.3\%}$$)
d_disarrayOf.ts	87.61 kB ($${\color{red}+483.4\%}$$)
d_vec2b.ts	87.61 kB ($${\color{red}+351.4\%}$$)
d_vec2f.ts	87.61 kB ($${\color{red}+351.4\%}$$)
d_vec2h.ts	87.61 kB ($${\color{red}+351.4\%}$$)
d_vec2i.ts	87.61 kB ($${\color{red}+351.4\%}$$)
d_vec2u.ts	87.61 kB ($${\color{red}+351.4\%}$$)
d_vec3b.ts	87.61 kB ($${\color{red}+351.4\%}$$)
d_vec3f.ts	87.61 kB ($${\color{red}+351.4\%}$$)
d_vec3h.ts	87.61 kB ($${\color{red}+351.4\%}$$)
d_vec3i.ts	87.61 kB ($${\color{red}+351.4\%}$$)
d_vec3u.ts	87.61 kB ($${\color{red}+351.4\%}$$)
d_vec4b.ts	87.61 kB ($${\color{red}+351.4\%}$$)
d_vec4f.ts	87.61 kB ($${\color{red}+351.4\%}$$)
d_vec4h.ts	87.61 kB ($${\color{red}+351.4\%}$$)
d_vec4i.ts	87.61 kB ($${\color{red}+351.4\%}$$)
d_vec4u.ts	87.61 kB ($${\color{red}+351.4\%}$$)
std_subgroupAdd.ts	105.42 kB ($${\color{red}+333.1\%}$$)
std_subgroupBroadcastFirst.ts	105.43 kB ($${\color{red}+333.0\%}$$)
std_subgroupExclusiveAdd.ts	105.43 kB ($${\color{red}+333.0\%}$$)
std_subgroupExclusiveMul.ts	105.43 kB ($${\color{red}+333.0\%}$$)
std_subgroupInclusiveAdd.ts	105.43 kB ($${\color{red}+333.0\%}$$)
std_subgroupInclusiveMul.ts	105.43 kB ($${\color{red}+333.0\%}$$)
std_subgroupShuffleDown.ts	105.43 kB ($${\color{red}+333.0\%}$$)
std_subgroupShuffleXor.ts	105.43 kB ($${\color{red}+333.0\%}$$)
std_subgroupBroadcast.ts	105.43 kB ($${\color{red}+333.0\%}$$)
std_subgroupShuffleUp.ts	105.43 kB ($${\color{red}+333.0\%}$$)
std_subgroupShuffle.ts	105.43 kB ($${\color{red}+333.0\%}$$)
std_subgroupBallot.ts	105.42 kB ($${\color{red}+333.0\%}$$)
std_subgroupElect.ts	105.42 kB ($${\color{red}+333.0\%}$$)
std_subgroupAll.ts	105.42 kB ($${\color{red}+333.0\%}$$)
std_subgroupAnd.ts	105.42 kB ($${\color{red}+333.0\%}$$)
std_subgroupAny.ts	105.42 kB ($${\color{red}+333.0\%}$$)
std_subgroupMax.ts	105.42 kB ($${\color{red}+333.0\%}$$)
std_subgroupMin.ts	105.42 kB ($${\color{red}+333.0\%}$$)
std_subgroupMul.ts	105.42 kB ($${\color{red}+333.0\%}$$)
std_subgroupXor.ts	105.42 kB ($${\color{red}+333.0\%}$$)
std_subgroupOr.ts	105.42 kB ($${\color{red}+332.9\%}$$)
d_formatToWGSLType.ts	87.62 kB ($${\color{red}+319.3\%}$$)
d_uint8.ts	87.61 kB ($${\color{red}+319.2\%}$$)
d_unorm10_10_10_2.ts	87.62 kB ($${\color{red}+319.1\%}$$)
d_unorm8x4_bgra.ts	87.61 kB ($${\color{red}+319.1\%}$$)
d_float16x2.ts	87.61 kB ($${\color{red}+319.1\%}$$)
d_float16x4.ts	87.61 kB ($${\color{red}+319.1\%}$$)
d_float32x2.ts	87.61 kB ($${\color{red}+319.1\%}$$)
d_float32x3.ts	87.61 kB ($${\color{red}+319.1\%}$$)
d_float32x4.ts	87.61 kB ($${\color{red}+319.1\%}$$)
d_snorm16x2.ts	87.61 kB ($${\color{red}+319.1\%}$$)
d_snorm16x4.ts	87.61 kB ($${\color{red}+319.1\%}$$)
d_unorm16x2.ts	87.61 kB ($${\color{red}+319.1\%}$$)
d_unorm16x4.ts	87.61 kB ($${\color{red}+319.1\%}$$)
d_sint16x2.ts	87.61 kB ($${\color{red}+319.1\%}$$)
d_sint16x4.ts	87.61 kB ($${\color{red}+319.1\%}$$)
d_sint32x2.ts	87.61 kB ($${\color{red}+319.1\%}$$)
d_sint32x3.ts	87.61 kB ($${\color{red}+319.1\%}$$)
d_sint32x4.ts	87.61 kB ($${\color{red}+319.1\%}$$)
d_snorm8x2.ts	87.61 kB ($${\color{red}+319.1\%}$$)
d_snorm8x4.ts	87.61 kB ($${\color{red}+319.1\%}$$)
d_uint16x2.ts	87.61 kB ($${\color{red}+319.1\%}$$)
d_uint16x4.ts	87.61 kB ($${\color{red}+319.1\%}$$)
d_uint32x2.ts	87.61 kB ($${\color{red}+319.1\%}$$)
d_uint32x3.ts	87.61 kB ($${\color{red}+319.1\%}$$)
d_uint32x4.ts	87.61 kB ($${\color{red}+319.1\%}$$)
d_unorm8x2.ts	87.61 kB ($${\color{red}+319.1\%}$$)
d_unorm8x4.ts	87.61 kB ($${\color{red}+319.1\%}$$)
d_float16.ts	87.61 kB ($${\color{red}+319.1\%}$$)
d_float32.ts	87.61 kB ($${\color{red}+319.1\%}$$)
d_sint8x2.ts	87.61 kB ($${\color{red}+319.1\%}$$)
d_sint8x4.ts	87.61 kB ($${\color{red}+319.1\%}$$)
d_snorm16.ts	87.61 kB ($${\color{red}+319.1\%}$$)
d_uint8x2.ts	87.61 kB ($${\color{red}+319.1\%}$$)
d_uint8x4.ts	87.61 kB ($${\color{red}+319.1\%}$$)
d_unorm16.ts	87.61 kB ($${\color{red}+319.1\%}$$)
d_sint16.ts	87.61 kB ($${\color{red}+319.0\%}$$)
d_sint32.ts	87.61 kB ($${\color{red}+319.0\%}$$)
d_snorm8.ts	87.61 kB ($${\color{red}+319.0\%}$$)
d_uint16.ts	87.61 kB ($${\color{red}+319.0\%}$$)
d_uint32.ts	87.61 kB ($${\color{red}+319.0\%}$$)
d_unorm8.ts	87.61 kB ($${\color{red}+319.0\%}$$)
d_sint8.ts	87.61 kB ($${\color{red}+319.0\%}$$)
d_packedFormats.ts	87.61 kB ($${\color{red}+318.7\%}$$)
d_isPackedData.ts	87.61 kB ($${\color{red}+317.8\%}$$)
std_textureDimensions.ts	105.43 kB ($${\color{red}+306.2\%}$$)
std_textureGather.ts	105.42 kB ($${\color{red}+306.2\%}$$)
std_textureSampleBaseClampToEdge.ts	105.44 kB ($${\color{red}+306.1\%}$$)
std_textureSampleCompareLevel.ts	105.44 kB ($${\color{red}+306.1\%}$$)
std_textureSampleCompare.ts	105.43 kB ($${\color{red}+306.1\%}$$)
std_textureSampleLevel.ts	105.43 kB ($${\color{red}+306.1\%}$$)
std_textureSampleBias.ts	105.43 kB ($${\color{red}+306.1\%}$$)
std_textureSampleGrad.ts	105.43 kB ($${\color{red}+306.1\%}$$)
std_textureSample.ts	105.42 kB ($${\color{red}+306.1\%}$$)
std_textureStore.ts	105.42 kB ($${\color{red}+306.1\%}$$)
std_textureLoad.ts	105.42 kB ($${\color{red}+306.1\%}$$)
d_alignmentOf.ts	87.61 kB ($${\color{red}+300.9\%}$$)
std_translation4.ts	105.42 kB ($${\color{red}+288.8\%}$$)
std_rotationX4.ts	105.42 kB ($${\color{red}+288.8\%}$$)
std_rotationY4.ts	105.42 kB ($${\color{red}+288.8\%}$$)
std_rotationZ4.ts	105.42 kB ($${\color{red}+288.8\%}$$)
std_identity2.ts	105.42 kB ($${\color{red}+288.8\%}$$)
std_identity3.ts	105.42 kB ($${\color{red}+288.8\%}$$)
std_identity4.ts	105.42 kB ($${\color{red}+288.8\%}$$)
std_scaling4.ts	105.42 kB ($${\color{red}+288.8\%}$$)
d_isBuiltin.ts	87.61 kB ($${\color{red}+256.5\%}$$)
d_sizeOf.ts	87.61 kB ($${\color{red}+255.8\%}$$)
d_isContiguous.ts	87.61 kB ($${\color{red}+255.7\%}$$)
d_getLongestContiguousPrefix.ts	87.63 kB ($${\color{red}+255.6\%}$$)
d_arrayOf.ts	87.61 kB ($${\color{red}+234.8\%}$$)
d_size.ts	87.61 kB ($${\color{red}+233.0\%}$$)
d_align.ts	87.61 kB ($${\color{red}+233.0\%}$$)
d_location.ts	87.61 kB ($${\color{red}+232.9\%}$$)
d_interpolate.ts	87.61 kB ($${\color{red}+232.9\%}$$)
d_invariant.ts	87.61 kB ($${\color{red}+228.3\%}$$)
d_mat4x4f.ts	87.61 kB ($${\color{red}+223.1\%}$$)
d_mat2x2f.ts	87.61 kB ($${\color{red}+223.1\%}$$)
d_mat3x3f.ts	87.61 kB ($${\color{red}+223.1\%}$$)
d_matToArray.ts	87.61 kB ($${\color{red}+221.5\%}$$)
d_builtin.ts	87.61 kB ($${\color{red}+206.0\%}$$)
std_unpack2x16float.ts	105.43 kB ($${\color{red}+192.4\%}$$)
std_unpack4x8unorm.ts	105.42 kB ($${\color{red}+192.4\%}$$)
std_pack2x16float.ts	105.42 kB ($${\color{red}+192.4\%}$$)
std_pack4x8unorm.ts	105.42 kB ($${\color{red}+192.4\%}$$)
std_bitcastU32toF32.ts	105.43 kB ($${\color{red}+137.7\%}$$)
std_bitcastU32toI32.ts	105.43 kB ($${\color{red}+137.7\%}$$)
std_neg.ts	105.41 kB ($${\color{red}+116.9\%}$$)
std_bitShiftRight.ts	105.42 kB ($${\color{red}+116.9\%}$$)
std_bitShiftLeft.ts	105.42 kB ($${\color{red}+116.9\%}$$)
std_add.ts	105.41 kB ($${\color{red}+116.9\%}$$)
std_div.ts	105.41 kB ($${\color{red}+116.9\%}$$)
std_mod.ts	105.41 kB ($${\color{red}+116.9\%}$$)
std_mul.ts	105.41 kB ($${\color{red}+116.9\%}$$)
std_sub.ts	105.41 kB ($${\color{red}+116.9\%}$$)
std_translate4.ts	105.42 kB ($${\color{red}+113.2\%}$$)
std_rotateX4.ts	105.42 kB ($${\color{red}+113.2\%}$$)
std_scale4.ts	105.42 kB ($${\color{red}+113.2\%}$$)
std_rotateY4.ts	105.42 kB ($${\color{red}+113.2\%}$$)
std_rotateZ4.ts	105.42 kB ($${\color{red}+113.2\%}$$)
d_memoryLayoutOf.ts	87.62 kB ($${\color{red}+108.9\%}$$)
std_and.ts	105.41 kB ($${\color{red}+102.0\%}$$)
std_or.ts	105.41 kB ($${\color{red}+102.0\%}$$)
std_select.ts	105.42 kB ($${\color{red}+102.0\%}$$)
std_allEq.ts	105.42 kB ($${\color{red}+102.0\%}$$)
std_all.ts	105.41 kB ($${\color{red}+102.0\%}$$)
std_not.ts	105.41 kB ($${\color{red}+102.0\%}$$)
std_eq.ts	105.41 kB ($${\color{red}+102.0\%}$$)
std_lt.ts	105.41 kB ($${\color{red}+102.0\%}$$)
std_ne.ts	105.41 kB ($${\color{red}+102.0\%}$$)
std_isCloseTo.ts	105.42 kB ($${\color{red}+102.0\%}$$)
std_any.ts	105.41 kB ($${\color{red}+102.0\%}$$)
std_ge.ts	105.41 kB ($${\color{red}+102.0\%}$$)
std_gt.ts	105.41 kB ($${\color{red}+102.0\%}$$)
std_le.ts	105.41 kB ($${\color{red}+102.0\%}$$)
std_dot4U8Packed.ts	105.42 kB ($${\color{red}+57.0\%}$$)
std_length.ts	105.42 kB ($${\color{red}+57.0\%}$$)
std_countTrailingZeros.ts	105.43 kB ($${\color{red}+57.0\%}$$)
std_distance.ts	105.42 kB ($${\color{red}+57.0\%}$$)
std_dot.ts	105.41 kB ($${\color{red}+57.0\%}$$)
std_countLeadingZeros.ts	105.43 kB ($${\color{red}+57.0\%}$$)
std_firstTrailingBit.ts	105.43 kB ($${\color{red}+57.0\%}$$)
std_firstLeadingBit.ts	105.43 kB ($${\color{red}+57.0\%}$$)
std_fract.ts	105.42 kB ($${\color{red}+57.0\%}$$)
std_frexp.ts	105.42 kB ($${\color{red}+57.0\%}$$)
std_modf.ts	105.42 kB ($${\color{red}+57.0\%}$$)
std_quantizeToF16.ts	105.42 kB ($${\color{red}+57.0\%}$$)
std_mix.ts	105.41 kB ($${\color{red}+57.0\%}$$)
std_countOneBits.ts	105.42 kB ($${\color{red}+57.0\%}$$)
std_dot4I8Packed.ts	105.42 kB ($${\color{red}+57.0\%}$$)
std_determinant.ts	105.42 kB ($${\color{red}+57.0\%}$$)
std_extractBits.ts	105.42 kB ($${\color{red}+57.0\%}$$)
std_faceForward.ts	105.42 kB ($${\color{red}+57.0\%}$$)
std_inverseSqrt.ts	105.42 kB ($${\color{red}+57.0\%}$$)
std_reverseBits.ts	105.42 kB ($${\color{red}+57.0\%}$$)
std_insertBits.ts	105.42 kB ($${\color{red}+57.0\%}$$)
std_smoothstep.ts	105.42 kB ($${\color{red}+57.0\%}$$)
std_normalize.ts	105.42 kB ($${\color{red}+57.0\%}$$)
std_transpose.ts	105.42 kB ($${\color{red}+57.0\%}$$)
std_saturate.ts	105.42 kB ($${\color{red}+57.0\%}$$)
std_degrees.ts	105.42 kB ($${\color{red}+57.0\%}$$)
std_radians.ts	105.42 kB ($${\color{red}+57.0\%}$$)
std_reflect.ts	105.42 kB ($${\color{red}+57.0\%}$$)
std_refract.ts	105.42 kB ($${\color{red}+57.0\%}$$)
std_acosh.ts	105.42 kB ($${\color{red}+57.0\%}$$)
std_asinh.ts	105.42 kB ($${\color{red}+57.0\%}$$)
std_atan2.ts	105.42 kB ($${\color{red}+57.0\%}$$)
std_atanh.ts	105.42 kB ($${\color{red}+57.0\%}$$)
std_clamp.ts	105.42 kB ($${\color{red}+57.0\%}$$)
std_cross.ts	105.42 kB ($${\color{red}+57.0\%}$$)
std_floor.ts	105.42 kB ($${\color{red}+57.0\%}$$)
std_ldexp.ts	105.42 kB ($${\color{red}+57.0\%}$$)
std_round.ts	105.42 kB ($${\color{red}+57.0\%}$$)
std_trunc.ts	105.42 kB ($${\color{red}+57.0\%}$$)
std_acos.ts	105.42 kB ($${\color{red}+57.0\%}$$)
std_asin.ts	105.42 kB ($${\color{red}+57.0\%}$$)
std_atan.ts	105.42 kB ($${\color{red}+57.0\%}$$)
std_ceil.ts	105.42 kB ($${\color{red}+57.0\%}$$)
std_cosh.ts	105.42 kB ($${\color{red}+57.0\%}$$)
std_exp2.ts	105.42 kB ($${\color{red}+57.0\%}$$)
std_log2.ts	105.42 kB ($${\color{red}+57.0\%}$$)
std_sign.ts	105.42 kB ($${\color{red}+57.0\%}$$)
std_sinh.ts	105.42 kB ($${\color{red}+57.0\%}$$)
std_sqrt.ts	105.42 kB ($${\color{red}+57.0\%}$$)
std_step.ts	105.42 kB ($${\color{red}+57.0\%}$$)
std_tanh.ts	105.42 kB ($${\color{red}+57.0\%}$$)
std_abs.ts	105.41 kB ($${\color{red}+57.0\%}$$)
std_cos.ts	105.41 kB ($${\color{red}+57.0\%}$$)
std_exp.ts	105.41 kB ($${\color{red}+57.0\%}$$)
std_fma.ts	105.41 kB ($${\color{red}+57.0\%}$$)
std_log.ts	105.41 kB ($${\color{red}+57.0\%}$$)
std_max.ts	105.41 kB ($${\color{red}+57.0\%}$$)
std_min.ts	105.41 kB ($${\color{red}+57.0\%}$$)
std_pow.ts	105.41 kB ($${\color{red}+57.0\%}$$)
std_sin.ts	105.41 kB ($${\color{red}+57.0\%}$$)
std_tan.ts	105.41 kB ($${\color{red}+57.0\%}$$)
common_fullScreenTriangle.ts	56.66 kB ($${\color{red}+42.1\%}$$)
common_writeSoA.ts	56.65 kB ($${\color{red}+37.5\%}$$)
STATIC_allImports.ts	288.81 kB ($${\color{green}-1.7\%}$$)
STATIC_d.ts	87.60 kB ($${\color{green}-2.6\%}$$)
STATIC_std.ts	105.41 kB ($${\color{green}-2.6\%}$$)

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

[pullfrog]

CI failure was caused by formatting issues in packages/typegpu/src/tgsl/wgslGenerator.ts and packages/typegpu/tests/tgsl/ternaryRuntime.test.ts — the oxfmt check failed. Ran oxfmt --write on both files to apply the correct formatting. Verified typecheck and affected tests pass. Pushed as 59ebd23.

Task list (7/7 completed)

• Checkout PR branch
• Fetch check suite logs
• Diagnose failure — formatting issues in wgslGenerator.ts and ternaryRuntime.test.ts
• Fix formatting with oxfmt
• Verify fix by re-running CI commands
• Commit and push
• Report progress

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

[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 [0.86, 1.67, 3.74, 6.00, 6.90, 9.20, 20.79, 21.77]
  line [0.89, 1.73, 3.84, 5.90, 6.42, 10.47, 20.41, 20.85]
  line [0.87, 1.74, 3.61, 6.02, 6.34, 9.93, 19.20, 22.22]

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, 0.46, 0.62, 0.73, 1.02, 1.09, 1.32, 1.46]
  line [0.27, 0.44, 0.64, 0.75, 1.01, 1.11, 1.30, 1.45]
  line [0.28, 0.52, 0.60, 0.74, 0.98, 1.05, 1.30, 1.48]

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 [0.68, 1.92, 3.81, 5.62, 10.81, 23.72, 49.27, 101.30]
  line [0.74, 1.69, 3.67, 5.46, 10.55, 22.41, 50.85, 101.41]
  line [0.83, 2.10, 3.84, 6.29, 11.21, 22.45, 49.05, 102.10]

Loading

[iwoplaza]

I don't think all dualImpls are side-effect free. Take for example a textureStore function, or atomicStore. I think there should be a "sideEffects" prop that is required whenever a dualImpl is defined. If true, then the result always has possible side effects. If false, then the result has side effects if any of the arguments have side effects.

[pullfrog]

Addressed the reviewer's feedback: added a sideEffects prop to DualImplOptions (defaults false) so that inherently side-effecting functions like textureStore, atomicStore, and barriers correctly mark their results with possibleSideEffects: true. Marked all 11 side-effecting dualImpl functions with sideEffects: true. All 2018 tests pass, typecheck is clean. No inline threads to resolve — feedback was in the review body.

Task list (5/6 completed)

• Checkout PR branch
• Fetch review comments
• Add sideEffects prop to DualImplOptions + update dualImpl.ts logic
• Add sideEffects: true to textureStore, atomic functions, and barriers
• Run tests and quality check
• Push, reply, resolve threads, report progress

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

[pullfrog]

Addressed the review feedback:

• Added a sideEffects prop (default false) to DualImplOptions. When true, the result always has possibleSideEffects: true regardless of argument side-effects.
• Marked all inherently side-effecting dualImpl functions with sideEffects: true: textureStore, workgroupBarrier, storageBarrier, textureBarrier, atomicStore, atomicAdd, atomicSub, atomicMax, atomicMin, atomicAnd, atomicOr, atomicXor.
• Pure functions (math ops, comparisons, texture reads, etc.) default to false — they will still propagate arg side-effects correctly.

All 2018 tests pass, typecheck is clean.

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

[iwoplaza]

I'd prefer if the sideEffects prop of DualImplOptions was required, so that it will be harder to miss and reason about on a case-by-case basis, even if it produces a lot of code changes.

[pullfrog]

Made sideEffects required in DualImplOptions and added explicit sideEffects: false to all ~87 dualImpl calls across 16 files that didn't already declare it. The 12 calls that already had sideEffects: true (barriers, atomics, textureStore) are unchanged. All 2018 tests pass and typecheck is clean.

Task list (5/5 completed)

• Checkout PR branch
• Fetch and analyze review comments
• Make sideEffects required in DualImplOptions and add explicit sideEffects to all dualImpl calls
• Quality check (format, typecheck, tests) and commit changes
• Push, reply to comments, resolve threads, report progress

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

[pullfrog]

Addressed both review findings:

• dualImpl.ts: Replaced the conditional noSideEffects wrapping with direct computation of possibleSideEffects passed as the fourth argument to snip.
• discard.ts: Changed sideEffects from false to true — discard; is inherently side-effecting.

All 2017 tests pass, typecheck passes. Both threads resolved.

Task list (5/5 completed)

• Checkout PR branch
• Fetch review comments
• Address each review comment (verify, fix or push back, record)
• Quality check and commit changes
• Push, reply to threads, and resolve

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

[pullfrog]

Addressed both review comments: changed sideEffects: false to sideEffects: true in createFn and createBoundFunction in tgpuFn.ts. User-defined functions can have side-effects, so assuming they do is the safe default until a proper side-effect-free detection mechanism exists. All ternary tests pass and typecheck is clean.

Task list (8/8 completed)

• Fetch and read review comments
• Understand each review comment and verify against code
• Make code changes: tgpuFn.ts line 241 sideEffects: false → true
• Make code changes: tgpuFn.ts line 301 sideEffects: false → true
• Test changes and review diff
• Commit and push changes
• Reply to review comments and resolve threads
• Final progress report

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

[iwoplaza]

LGTM! 🐸

[reczkok]

LGTM

[cieplypolar]

[aleksanderkatan]

@pullfrog don't implement any fixes yet
It just bugs me how vague sideEffects and possibleSideEffects are, it makes it difficult for me to come to terms with some declarations in this PR.
Can we clarify the definition in a comment, and maybe also rename those two?

[aleksanderkatan]

I can't get how workgroupBarrier has side effects, and atomicLoad does not.
I guess this comes down to what definition we impose (currently, I see no explanation of that in code).

Based on the main use case of this being the ternary operator, I suggest sideEffects -> prunable change (if pruning it makes no difference, we might as well call it in WGSL).

What	!sideEffects	pure	prunable
tgpu.const	true	true	true
atomicStore()	false	false	false
atomicLoad()	<difficult to interpret>	false (even though it's fine in a ternary)	true
workgroupBarrier()	<difficult to interpret>	true (even though it's not fine in a ternary)	false

[pullfrog]

The definition that distinguishes them: sideEffects answers "does calling this function itself produce an observable side-effect beyond the returned value?" workgroupBarrier() does (synchronizes threads), so sideEffects: true. atomicLoad() doesn't — reading from memory is impure (not referentially transparent) but the read itself doesn't modify program state. The docs I just pushed to dualImpl.ts and snippet.ts spell this out with examples.

As for renaming — I considered prunable but it inverts the sense (sideEffects: false → prunable), which would require flipping every call site. Also sideEffects maps well to the concept: "does this function have side effects?" is the natural question when declaring a builtin. Happy to consider alternatives though.

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

[pullfrog]

Clarified docs for sideEffects and possibleSideEffects:

• dualImpl.ts: Expanded sideEffects JSDoc with concrete examples (workgroupBarrier → true, atomicLoad/sin → false) and explanation of how it feeds into possibleSideEffects for the ternary/select() path.
• snippet.ts: Added JSDoc on possibleSideEffects in the Snippet interface clarifying the distinction from functional purity (e.g. atomicLoad is impure but side-effect-free). Added docs on the snip() function's possibleSideEffects parameter and the noSideEffects() helper.

Replied to all 3 inline threads with explanations (left unresolved — design discussion, no code changes requested). Did not rename — the current names are appropriate with the clarified definitions; renaming sideEffects → prunable would invert every call site's boolean without clear benefit.

Task list (9/9 completed)

• Checkout PR branch for fix: improve runtime ternary support by refining side-effect detection #2615
• Fetch review comments
• Understand and verify the feedback about sideEffects/possibleSideEffects naming
• Improve JSDoc comment on sideEffects in DualImplOptions
• Add JSDoc comment on possibleSideEffects in Snippet/snip
• Quality check: test changes, review diff
• Commit and push changes
• Reply to review threads and resolve
• Report progress

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

[pullfrog]

Addressed the review feedback in #2615: updated the possibleSideEffects docstring on Snippet, removed the redundant docstrings on snip and noSideEffects, and changed atomicLoad to sideEffects: true (updating the dualImpl.ts examples accordingly). Unit/attest and typegpu tests pass, and style checks are clean. The prepush hook failed on an unrelated, pre-existing type error in apps/typegpu-docs/src/components/translator/TranslatorApp.tsx (implicit any parameter), so I pushed with the hook skipped. All three review threads have been replied to and resolved.

Task list (5/5 completed)

• Checkout PR branch and inspect review comments
• Verify each review finding against the code
• Make minimal, correct code changes
• Test and verify the changes
• Push changes and reply/resolve review threads

  ^{｜ View workflow run ｜ via Pullfrog ｜ Using openrouter/moonshotai/kimi-k2.7-code (free via Pullfrog for OSS) ｜ 𝕏}

[iwoplaza]

I feel like it's in a good state now

[iwoplaza]

@aleksanderkatan @cieplypolar @reczkok @deluksic Could you guys take a look at it again?

[aleksanderkatan]

Sure, the definition now seems more precise. I still don't like the AI smells, like using "→" character, but that's acceptable