Is there any interest in supporting (ASCII-only) case-insensitive substring search in memchr? If so, I'd be happy to propose a candidate API and do the implementation work.
Context / motivation: Apache DataFusion and arrow-rs use memchr to implement SQL constructs such as LIKE '%foo', with great performance (thanks for your work on memchr!). Another widely used SQL construct is ILIKE (case insensitive search), which currently has to be implemented via regex. If memchr was extended to support SIMD-accelerated case-insensitive search for ASCII strings, we'd be able to accelerate ILIKE significantly (roughly 2-3x over the regex-based implementation).
Is there any interest in supporting (ASCII-only) case-insensitive substring search in
memchr? If so, I'd be happy to propose a candidate API and do the implementation work.Context / motivation: Apache DataFusion and arrow-rs use
memchrto implement SQL constructs such asLIKE '%foo', with great performance (thanks for your work onmemchr!). Another widely used SQL construct isILIKE(case insensitive search), which currently has to be implemented via regex. Ifmemchrwas extended to support SIMD-accelerated case-insensitive search for ASCII strings, we'd be able to accelerateILIKEsignificantly (roughly 2-3x over the regex-based implementation).