You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Have questions about stdlib? Need help fixing a bug, figuring out what to do next, or just looking for feedback? Join our weekly office hours to connect with project maintainers, stay updated on the latest project news, and chat with other community members. This is a great opportunity to ask questions, share ideas, and engage directly with the stdlib team.
I had a doubt regarding find, findIndex, and filter benchmarks for array/uint64 and want to confirm the convention.
array/uint32 uses a zero-filled array with a v > 0 predicate, so nothing matches and the full scan is measured; it also adds a separate ::this_context: benchmark for the thisArg overload.
array/complex128 uses a predicate that matches the first element, so it returns immediately and doesn't measure the scan, with no thisArg benchmark.
Should number/int32/base/signbit check if the provided value is a 32-bit signed integer?
Discuss Issue #1088 and my approach( I鈥檝e outlined my proposed approach in the issue comments and would appreciate any feedback before moving forward )
I'd like to discuss stdlib#15106, particularly the choice of Highway, the reusable native runtime build, and the opt-in setup with SIMD_BACKEND=highway.
Pre-existing test Failures(dsort): I've encountered some test failures in blas/ext/sort/test/test.js while working on my PR for blas/ext/base/ndarray/dsort(#14957). I've verified these failures are pre-existing on develop ;. They stem from a gap in nullary-strided1d-dispatch-factory, which doesn't throw an error when a broadcast-incompatible sortOrder (e.g., shape [4] or [2,2,2]) is passed with a generic dtype input of shape [2,2] when using gsort as a fallback. Should I fix this as part of my current PR?
Guidance on PR #14921: I would appreciate some guidance on my current approach for C implementation ;
Activity
DivitJain26 commentedon Sep 8, 2026
I had a doubt regarding
find,findIndex, andfilterbenchmarks forarray/uint64and want to confirm the convention.array/uint32uses a zero-filled array with av > 0predicate, so nothing matches and the full scan is measured; it also adds a separate::this_context:benchmark for thethisArgoverload.array/complex128uses a predicate that matches the first element, so it returns immediately and doesn't measure the scan, with nothisArgbenchmark.Should
number/int32/base/signbitcheck if the provided value is a 32-bit signed integer?Discussion on
has-equal-values.0PrashantYadav0 commentedon Sep 8, 2026
Discuss Issue #1088 and my approach( I鈥檝e outlined my proposed approach in the issue comments and would appreciate any feedback before moving forward )
teddygood commentedon Sep 8, 2026
I'd like to discuss stdlib#15106, particularly the choice of Highway, the reusable native runtime build, and the opt-in setup with
SIMD_BACKEND=highway.manit2004 commentedon Sep 8, 2026
Request review on #14415
Discuss about stringarray implementation
ujjwalv01 commentedon Sep 8, 2026
Pre-existing test Failures(
dsort): I've encountered some test failures inblas/ext/sort/test/test.jswhile working on my PR forblas/ext/base/ndarray/dsort(#14957). I've verified these failures are pre-existing ondevelop;. They stem from a gap innullary-strided1d-dispatch-factory, which doesn't throw an error when a broadcast-incompatiblesortOrder(e.g., shape[4]or[2,2,2]) is passed with agenericdtype input of shape[2,2]when usinggsortas a fallback. Should I fix this as part of my current PR?Guidance on PR #14921: I would appreciate some guidance on my current approach for C implementation ;