Add ASCII fast path for ILIKE scalar (90% faster) #3306

tustvold · 2022-12-09T12:17:51Z

Which issue does this PR close?

Rationale for this change

ilike_utf8 scalar equals
                        time:   [218.99 µs 219.05 µs 219.10 µs]
                        change: [-90.262% -90.257% -90.253%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 9 outliers among 100 measurements (9.00%)
  5 (5.00%) high mild
  4 (4.00%) high severe

Benchmarking ilike_utf8 scalar contains: Warming up for 3.0000 s
Warning: Unable to complete 100 samples in 5.0s. You may wish to increase target time to 9.1s, enable flat sampling, or reduce sample count to 50.
ilike_utf8 scalar contains
                        time:   [1.8097 ms 1.8104 ms 1.8111 ms]
                        change: [-56.369% -56.326% -56.264%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 9 outliers among 100 measurements (9.00%)
  3 (3.00%) low mild
  4 (4.00%) high mild
  2 (2.00%) high severe

ilike_utf8 scalar ends with
                        time:   [261.52 µs 261.58 µs 261.65 µs]
                        change: [-88.863% -88.850% -88.831%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 4 outliers among 100 measurements (4.00%)
  1 (1.00%) high mild
  3 (3.00%) high severe

ilike_utf8 scalar starts with
                        time:   [265.40 µs 265.45 µs 265.50 µs]
                        change: [-88.692% -88.679% -88.656%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 11 outliers among 100 measurements (11.00%)
  2 (2.00%) low mild
  6 (6.00%) high mild
  3 (3.00%) high severe

Benchmarking ilike_utf8 scalar complex: Warming up for 3.0000 s
Warning: Unable to complete 100 samples in 5.0s. You may wish to increase target time to 9.2s, enable flat sampling, or reduce sample count to 50.
ilike_utf8 scalar complex
                        time:   [1.8319 ms 1.8327 ms 1.8334 ms]
                        change: [-4.2733% -4.1030% -3.9344%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 4 outliers among 100 measurements (4.00%)
  3 (3.00%) high mild
  1 (1.00%) high severe

nilike_utf8 scalar equals
                        time:   [232.53 µs 232.59 µs 232.65 µs]
                        change: [-90.314% -90.298% -90.265%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 8 outliers among 100 measurements (8.00%)
  5 (5.00%) high mild
  3 (3.00%) high severe

Benchmarking nilike_utf8 scalar contains: Warming up for 3.0000 s
Warning: Unable to complete 100 samples in 5.0s. You may wish to increase target time to 9.1s, enable flat sampling, or reduce sample count to 50.
nilike_utf8 scalar contains
                        time:   [1.7942 ms 1.7962 ms 1.7993 ms]
                        change: [-57.816% -57.765% -57.701%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 6 outliers among 100 measurements (6.00%)
  3 (3.00%) high mild
  3 (3.00%) high severe

nilike_utf8 scalar ends with
                        time:   [243.52 µs 243.58 µs 243.66 µs]
                        change: [-89.761% -89.758% -89.754%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 9 outliers among 100 measurements (9.00%)
  1 (1.00%) low mild
  7 (7.00%) high mild
  1 (1.00%) high severe

nilike_utf8 scalar starts with
                        time:   [253.10 µs 253.18 µs 253.28 µs]
                        change: [-89.367% -89.347% -89.313%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 6 outliers among 100 measurements (6.00%)
  1 (1.00%) high mild
  5 (5.00%) high severe

Benchmarking nilike_utf8 scalar complex: Warming up for 3.0000 s
Warning: Unable to complete 100 samples in 5.0s. You may wish to increase target time to 9.2s, enable flat sampling, or reduce sample count to 50.
nilike_utf8 scalar complex
                        time:   [1.8143 ms 1.8152 ms 1.8162 ms]
                        change: [-5.6406% -5.5148% -5.3303%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 9 outliers among 100 measurements (9.00%)
  4 (4.00%) high mild
  5 (5.00%) high severe

What changes are included in this PR?

Are there any user-facing changes?

tustvold · 2022-12-09T12:20:58Z

FYI @askoa

alamb · 2022-12-09T14:56:48Z

arrow-string/src/like.rs

-    } else if right.starts_with('%') && !right[1..].contains(is_like_pattern) {
-        // fast path, can use ends_with
-        let ends_str = &right[1..].to_uppercase();
+    // If not ASCII faster to use case insensitive regex than allocating using to_uppercase


Suggested change

// If not ASCII faster to use case insensitive regex than allocating using to_uppercase

// If ASCII faster to use case insensitive regex than allocating using to_uppercase

alamb

Do the existing tests cover all the corner cases?

tustvold · 2022-12-09T15:01:20Z

I can add some more, give me a minute

tustvold · 2022-12-09T15:04:09Z

I think this might not work for non-ASCII characters that capitalize to ASCII characters 😢

alamb · 2022-12-09T15:23:17Z

Perhaps you can also check that left is entirely ASCII (which is a fairly common case)

tustvold · 2022-12-09T15:30:33Z

Interestingly the regex crate doesn't appear to handle this correctly... 🤔

tustvold · 2022-12-09T16:00:36Z

I think this might not work for non-ASCII characters that capitalize to ASCII characters

So it turns out the fact we were handling this was actually incorrect, so this is not only faster, but more correct 😆 - #3311

tustvold · 2022-12-09T16:17:14Z

Aargh, there are unicode characters that have lowercase ASCII... Guess I will need to check if the string are ASCII, and use the regex if not

tustvold · 2022-12-09T17:20:54Z

This is currently blocked on #3313, which will allow the kernel to assume a GenericStringArray, which in turn will allow ASCII verification in one pass.

tustvold · 2022-12-09T20:34:04Z

ilike_utf8 scalar equals
                        time:   [271.37 µs 271.48 µs 271.62 µs]
                        change: [-87.932% -87.924% -87.914%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 13 outliers among 100 measurements (13.00%)
  2 (2.00%) low mild
  4 (4.00%) high mild
  7 (7.00%) high severe

Benchmarking ilike_utf8 scalar contains: Warming up for 3.0000 s
Warning: Unable to complete 100 samples in 5.0s. You may wish to increase target time to 9.1s, enable flat sampling, or reduce sample count to 50.
ilike_utf8 scalar contains
                        time:   [1.8028 ms 1.8037 ms 1.8048 ms]
                        change: [-56.517% -56.488% -56.455%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 10 outliers among 100 measurements (10.00%)
  5 (5.00%) high mild
  5 (5.00%) high severe

ilike_utf8 scalar ends with
                        time:   [287.38 µs 287.52 µs 287.65 µs]
                        change: [-87.759% -87.745% -87.729%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 3 outliers among 100 measurements (3.00%)
  3 (3.00%) high severe

ilike_utf8 scalar starts with
                        time:   [273.73 µs 273.85 µs 273.98 µs]
                        change: [-88.337% -88.328% -88.318%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 11 outliers among 100 measurements (11.00%)
  1 (1.00%) low mild
  7 (7.00%) high mild
  3 (3.00%) high severe

Benchmarking ilike_utf8 scalar complex: Warming up for 3.0000 s
Warning: Unable to complete 100 samples in 5.0s. You may wish to increase target time to 9.3s, enable flat sampling, or reduce sample count to 50.
ilike_utf8 scalar complex
                        time:   [1.8338 ms 1.8349 ms 1.8361 ms]
                        change: [-4.1219% -3.9358% -3.7411%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 9 outliers among 100 measurements (9.00%)
  4 (4.00%) high mild
  5 (5.00%) high severe

nilike_utf8 scalar equals
                        time:   [257.80 µs 257.85 µs 257.92 µs]
                        change: [-89.265% -89.246% -89.211%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 9 outliers among 100 measurements (9.00%)
  4 (4.00%) high mild
  5 (5.00%) high severe

Benchmarking nilike_utf8 scalar contains: Warming up for 3.0000 s
Warning: Unable to complete 100 samples in 5.0s. You may wish to increase target time to 9.2s, enable flat sampling, or reduce sample count to 50.
nilike_utf8 scalar contains
                        time:   [1.8283 ms 1.8293 ms 1.8303 ms]
                        change: [-57.048% -56.996% -56.926%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 6 outliers among 100 measurements (6.00%)
  2 (2.00%) high mild
  4 (4.00%) high severe

nilike_utf8 scalar ends with
                        time:   [256.38 µs 256.46 µs 256.54 µs]
                        change: [-89.212% -89.192% -89.160%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 8 outliers among 100 measurements (8.00%)
  4 (4.00%) high mild
  4 (4.00%) high severe

nilike_utf8 scalar starts with
                        time:   [262.65 µs 262.70 µs 262.77 µs]
                        change: [-88.974% -88.958% -88.927%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 4 outliers among 100 measurements (4.00%)
  1 (1.00%) high mild
  3 (3.00%) high severe

Benchmarking nilike_utf8 scalar complex: Warming up for 3.0000 s
Warning: Unable to complete 100 samples in 5.0s. You may wish to increase target time to 9.3s, enable flat sampling, or reduce sample count to 50.
nilike_utf8 scalar complex
                        time:   [1.8537 ms 1.8547 ms 1.8559 ms]
                        change: [-3.5514% -3.4042% -3.2212%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 13 outliers among 100 measurements (13.00%)
  2 (2.00%) low mild
  3 (3.00%) high mild
  8 (8.00%) high severe

ilike_utf8_scalar_dyn dictionary[10] string[4])
                        time:   [59.817 µs 59.887 µs 59.968 µs]
                        change: [-97.558% -97.553% -97.546%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 12 outliers among 100 measurements (12.00%)
  5 (5.00%) high mild
  7 (7.00%) high severe

So even with the extra check to detect is_ascii, this still is a pretty significant speedup

alamb · 2022-12-09T21:33:01Z

arrow-string/src/like.rs

@@ -1272,6 +1270,101 @@ mod tests {
        vec![true, false, false, false]
    );

+    // We only implement loose matching


When I ran these tests without the changes in this PR I got a bunch of failures. Is that expected?

failures: ---- like::tests::test_utf8_array_ilike_unicode stdout ---- thread 'like::tests::test_utf8_array_ilike_unicode' panicked at 'assertion failed: `(left == right)` left: `true`, right: `false`: unexpected result when comparing FFkoß at position 0 to FFkoSS ', arrow-string/src/like.rs:1279:5 stack backtrace: 0: rust_begin_unwind at /rustc/897e37553bba8b42751c67658967889d11ecd120/library/std/src/panicking.rs:584:5 1: core::panicking::panic_fmt at /rustc/897e37553bba8b42751c67658967889d11ecd120/library/core/src/panicking.rs:142:14 2: core::panicking::assert_failed_inner at /rustc/897e37553bba8b42751c67658967889d11ecd120/library/core/src/panicking.rs:218:23 3: core::panicking::assert_failed at /rustc/897e37553bba8b42751c67658967889d11ecd120/library/core/src/panicking.rs:181:5 4: arrow_string::like::tests::test_utf8_array_ilike_unicode at ./src/like.rs:1279:5 5: arrow_string::like::tests::test_utf8_array_ilike_unicode::{{closure}} at ./src/like.rs:917:13 6: core::ops::function::FnOnce::call_once at /rustc/897e37553bba8b42751c67658967889d11ecd120/library/core/src/ops/function.rs:248:5 7: core::ops::function::FnOnce::call_once at /rustc/897e37553bba8b42751c67658967889d11ecd120/library/core/src/ops/function.rs:248:5 note: Some details are omitted, run with `RUST_BACKTRACE=full` for a verbose backtrace. ---- like::tests::test_utf8_array_ilike_unicode_contains stdout ---- thread 'like::tests::test_utf8_array_ilike_unicode_contains' panicked at 'assertion failed: `(left == right)` left: `true`, right: `false`: unexpected result when comparing sdlkdfFkoßsdfs at position 0 to %FFkoSS% ', arrow-string/src/like.rs:1331:5 stack backtrace: 0: rust_begin_unwind at /rustc/897e37553bba8b42751c67658967889d11ecd120/library/std/src/panicking.rs:584:5 1: core::panicking::panic_fmt at /rustc/897e37553bba8b42751c67658967889d11ecd120/library/core/src/panicking.rs:142:14 2: core::panicking::assert_failed_inner at /rustc/897e37553bba8b42751c67658967889d11ecd120/library/core/src/panicking.rs:218:23 3: core::panicking::assert_failed at /rustc/897e37553bba8b42751c67658967889d11ecd120/library/core/src/panicking.rs:181:5 4: arrow_string::like::tests::test_utf8_array_ilike_unicode_contains at ./src/like.rs:1331:5 5: arrow_string::like::tests::test_utf8_array_ilike_unicode_contains::{{closure}} at ./src/like.rs:917:13 6: core::ops::function::FnOnce::call_once at /rustc/897e37553bba8b42751c67658967889d11ecd120/library/core/src/ops/function.rs:248:5 7: core::ops::function::FnOnce::call_once at /rustc/897e37553bba8b42751c67658967889d11ecd120/library/core/src/ops/function.rs:248:5 note: Some details are omitted, run with `RUST_BACKTRACE=full` for a verbose backtrace. ---- like::tests::test_utf8_array_ilike_unicode_contains_dyn stdout ---- thread 'like::tests::test_utf8_array_ilike_unicode_contains_dyn' panicked at 'assertion failed: `(left == right)` left: `true`, right: `false`: unexpected result when comparing sdlkdfFkoßsdfs at position 0 to %FFkoSS% ', arrow-string/src/like.rs:1331:5 stack backtrace: 0: rust_begin_unwind at /rustc/897e37553bba8b42751c67658967889d11ecd120/library/std/src/panicking.rs:584:5 1: core::panicking::panic_fmt at /rustc/897e37553bba8b42751c67658967889d11ecd120/library/core/src/panicking.rs:142:14 2: core::panicking::assert_failed_inner at /rustc/897e37553bba8b42751c67658967889d11ecd120/library/core/src/panicking.rs:218:23 3: core::panicking::assert_failed at /rustc/897e37553bba8b42751c67658967889d11ecd120/library/core/src/panicking.rs:181:5 4: arrow_string::like::tests::test_utf8_array_ilike_unicode_contains_dyn at ./src/like.rs:1331:5 5: arrow_string::like::tests::test_utf8_array_ilike_unicode_contains_dyn::{{closure}} at ./src/like.rs:917:13 6: core::ops::function::FnOnce::call_once at /rustc/897e37553bba8b42751c67658967889d11ecd120/library/core/src/ops/function.rs:248:5 7: core::ops::function::FnOnce::call_once at /rustc/897e37553bba8b42751c67658967889d11ecd120/library/core/src/ops/function.rs:248:5 note: Some details are omitted, run with `RUST_BACKTRACE=full` for a verbose backtrace. ---- like::tests::test_utf8_array_ilike_unicode_dyn stdout ---- thread 'like::tests::test_utf8_array_ilike_unicode_dyn' panicked at 'assertion failed: `(left == right)` left: `true`, right: `false`: unexpected result when comparing FFkoß at position 0 to FFkoSS ', arrow-string/src/like.rs:1279:5 stack backtrace: 0: rust_begin_unwind at /rustc/897e37553bba8b42751c67658967889d11ecd120/library/std/src/panicking.rs:584:5 1: core::panicking::panic_fmt at /rustc/897e37553bba8b42751c67658967889d11ecd120/library/core/src/panicking.rs:142:14 2: core::panicking::assert_failed_inner at /rustc/897e37553bba8b42751c67658967889d11ecd120/library/core/src/panicking.rs:218:23 3: core::panicking::assert_failed at /rustc/897e37553bba8b42751c67658967889d11ecd120/library/core/src/panicking.rs:181:5 4: arrow_string::like::tests::test_utf8_array_ilike_unicode_dyn at ./src/like.rs:1279:5 5: arrow_string::like::tests::test_utf8_array_ilike_unicode_dyn::{{closure}} at ./src/like.rs:917:13 6: core::ops::function::FnOnce::call_once at /rustc/897e37553bba8b42751c67658967889d11ecd120/library/core/src/ops/function.rs:248:5 7: core::ops::function::FnOnce::call_once at /rustc/897e37553bba8b42751c67658967889d11ecd120/library/core/src/ops/function.rs:248:5 note: Some details are omitted, run with `RUST_BACKTRACE=full` for a verbose backtrace. ---- like::tests::test_utf8_array_ilike_unicode_ends stdout ---- thread 'like::tests::test_utf8_array_ilike_unicode_ends' panicked at 'assertion failed: `(left == right)` left: `true`, right: `false`: unexpected result when comparing sdlkdfFFkoß at position 0 to %FFkoSS ', arrow-string/src/like.rs:1311:5 stack backtrace: 0: rust_begin_unwind at /rustc/897e37553bba8b42751c67658967889d11ecd120/library/std/src/panicking.rs:584:5 1: core::panicking::panic_fmt at /rustc/897e37553bba8b42751c67658967889d11ecd120/library/core/src/panicking.rs:142:14 2: core::panicking::assert_failed_inner at /rustc/897e37553bba8b42751c67658967889d11ecd120/library/core/src/panicking.rs:218:23 3: core::panicking::assert_failed at /rustc/897e37553bba8b42751c67658967889d11ecd120/library/core/src/panicking.rs:181:5 4: arrow_string::like::tests::test_utf8_array_ilike_unicode_ends at ./src/like.rs:924:21 5: arrow_string::like::tests::test_utf8_array_ilike_unicode_ends::{{closure}} at ./src/like.rs:917:13 6: core::ops::function::FnOnce::call_once at /rustc/897e37553bba8b42751c67658967889d11ecd120/library/core/src/ops/function.rs:248:5 7: core::ops::function::FnOnce::call_once at /rustc/897e37553bba8b42751c67658967889d11ecd120/library/core/src/ops/function.rs:248:5 note: Some details are omitted, run with `RUST_BACKTRACE=full` for a verbose backtrace. ---- like::tests::test_utf8_array_ilike_unicode_ends_dyn stdout ---- thread 'like::tests::test_utf8_array_ilike_unicode_ends_dyn' panicked at 'assertion failed: `(left == right)` left: `true`, right: `false`: unexpected result when comparing sdlkdfFFkoß at position 0 to %FFkoSS ', arrow-string/src/like.rs:1311:5 stack backtrace: 0: rust_begin_unwind at /rustc/897e37553bba8b42751c67658967889d11ecd120/library/std/src/panicking.rs:584:5 1: core::panicking::panic_fmt at /rustc/897e37553bba8b42751c67658967889d11ecd120/library/core/src/panicking.rs:142:14 2: core::panicking::assert_failed_inner at /rustc/897e37553bba8b42751c67658967889d11ecd120/library/core/src/panicking.rs:218:23 3: core::panicking::assert_failed at /rustc/897e37553bba8b42751c67658967889d11ecd120/library/core/src/panicking.rs:181:5 4: arrow_string::like::tests::test_utf8_array_ilike_unicode_ends_dyn at ./src/like.rs:924:21 5: arrow_string::like::tests::test_utf8_array_ilike_unicode_ends_dyn::{{closure}} at ./src/like.rs:917:13 6: core::ops::function::FnOnce::call_once at /rustc/897e37553bba8b42751c67658967889d11ecd120/library/core/src/ops/function.rs:248:5 7: core::ops::function::FnOnce::call_once at /rustc/897e37553bba8b42751c67658967889d11ecd120/library/core/src/ops/function.rs:248:5 note: Some details are omitted, run with `RUST_BACKTRACE=full` for a verbose backtrace. ---- like::tests::test_utf8_array_ilike_unicode_starts stdout ---- thread 'like::tests::test_utf8_array_ilike_unicode_starts' panicked at 'assertion failed: `(left == right)` left: `true`, right: `false`: unexpected result when comparing FFkoßsdlkdf at position 0 to FFkoSS% ', arrow-string/src/like.rs:1291:5 stack backtrace: 0: rust_begin_unwind at /rustc/897e37553bba8b42751c67658967889d11ecd120/library/std/src/panicking.rs:584:5 1: core::panicking::panic_fmt at /rustc/897e37553bba8b42751c67658967889d11ecd120/library/core/src/panicking.rs:142:14 2: core::panicking::assert_failed_inner at /rustc/897e37553bba8b42751c67658967889d11ecd120/library/core/src/panicking.rs:218:23 3: core::panicking::assert_failed at /rustc/897e37553bba8b42751c67658967889d11ecd120/library/core/src/panicking.rs:181:5 4: arrow_string::like::tests::test_utf8_array_ilike_unicode_starts at ./src/like.rs:1291:5 5: arrow_string::like::tests::test_utf8_array_ilike_unicode_starts::{{closure}} at ./src/like.rs:917:13 6: core::ops::function::FnOnce::call_once at /rustc/897e37553bba8b42751c67658967889d11ecd120/library/core/src/ops/function.rs:248:5 7: core::ops::function::FnOnce::call_once at /rustc/897e37553bba8b42751c67658967889d11ecd120/library/core/src/ops/function.rs:248:5 note: Some details are omitted, run with `RUST_BACKTRACE=full` for a verbose backtrace. ---- like::tests::test_utf8_array_ilike_unicode_start_dyn stdout ---- thread 'like::tests::test_utf8_array_ilike_unicode_start_dyn' panicked at 'assertion failed: `(left == right)` left: `true`, right: `false`: unexpected result when comparing FFkoßsdlkdf at position 0 to FFkoSS% ', arrow-string/src/like.rs:1291:5 stack backtrace: 0: rust_begin_unwind at /rustc/897e37553bba8b42751c67658967889d11ecd120/library/std/src/panicking.rs:584:5 1: core::panicking::panic_fmt at /rustc/897e37553bba8b42751c67658967889d11ecd120/library/core/src/panicking.rs:142:14 2: core::panicking::assert_failed_inner at /rustc/897e37553bba8b42751c67658967889d11ecd120/library/core/src/panicking.rs:218:23 3: core::panicking::assert_failed at /rustc/897e37553bba8b42751c67658967889d11ecd120/library/core/src/panicking.rs:181:5 4: arrow_string::like::tests::test_utf8_array_ilike_unicode_start_dyn at ./src/like.rs:1291:5 5: arrow_string::like::tests::test_utf8_array_ilike_unicode_start_dyn::{{closure}} at ./src/like.rs:917:13 6: core::ops::function::FnOnce::call_once at /rustc/897e37553bba8b42751c67658967889d11ecd120/library/core/src/ops/function.rs:248:5 7: core::ops::function::FnOnce::call_once at /rustc/897e37553bba8b42751c67658967889d11ecd120/library/core/src/ops/function.rs:248:5 note: Some details are omitted, run with `RUST_BACKTRACE=full` for a verbose backtrace. failures: like::tests::test_utf8_array_ilike_unicode like::tests::test_utf8_array_ilike_unicode_contains like::tests::test_utf8_array_ilike_unicode_contains_dyn like::tests::test_utf8_array_ilike_unicode_dyn like::tests::test_utf8_array_ilike_unicode_ends like::tests::test_utf8_array_ilike_unicode_ends_dyn like::tests::test_utf8_array_ilike_unicode_start_dyn like::tests::test_utf8_array_ilike_unicode_starts

Yes see #3311 the previous special case logic was incorrect as it changed the semantics from the regex

alamb

LGTM -- cc @Dandandan

tustvold added 2 commits December 9, 2022 11:53

Add ASCII fast path for ILIKE scalar

0fb5ea5

Update starts_with and ends_with

616342d

github-actions bot added the arrow Changes to the arrow crate label Dec 9, 2022

tustvold force-pushed the ilike-fast-path branch 2 times, most recently from 5d73cc4 to 160ee74 Compare December 9, 2022 12:18

Avoid allocations

faf6eaa

tustvold force-pushed the ilike-fast-path branch from 160ee74 to faf6eaa Compare December 9, 2022 12:19

Format

a47500a

tustvold requested a review from viirya December 9, 2022 12:24

alamb reviewed Dec 9, 2022

View reviewed changes

tustvold marked this pull request as draft December 9, 2022 15:06

Add more tests

7b2c61b

tustvold force-pushed the ilike-fast-path branch from 4597713 to 7b2c61b Compare December 9, 2022 15:59

tustvold marked this pull request as ready for review December 9, 2022 16:01

tustvold marked this pull request as draft December 9, 2022 16:21

Add failing test

3ea923e

tustvold added 2 commits December 9, 2022 20:11

Merge remote-tracking branch 'upstream/master' into ilike-fast-path

3d5793f

Verify is_ascii

3ff4180

tustvold marked this pull request as ready for review December 9, 2022 20:34

alamb reviewed Dec 9, 2022

View reviewed changes

alamb approved these changes Dec 9, 2022

View reviewed changes

tustvold merged commit 9e39f96 into apache:master Dec 9, 2022

alamb mentioned this pull request Dec 29, 2022

ILIKE Kernels Inconsistent Case Folding #3311

Closed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Add ASCII fast path for ILIKE scalar (90% faster) #3306

Add ASCII fast path for ILIKE scalar (90% faster) #3306

tustvold commented Dec 9, 2022 •

edited

Loading

tustvold commented Dec 9, 2022

alamb Dec 9, 2022

tustvold Dec 9, 2022

alamb left a comment

tustvold commented Dec 9, 2022

tustvold commented Dec 9, 2022

alamb commented Dec 9, 2022

tustvold commented Dec 9, 2022

tustvold commented Dec 9, 2022

tustvold commented Dec 9, 2022 •

edited

Loading

tustvold commented Dec 9, 2022

tustvold commented Dec 9, 2022

alamb Dec 9, 2022

tustvold Dec 9, 2022 •

edited

Loading

alamb left a comment

	// If not ASCII faster to use case insensitive regex than allocating using to_uppercase
	// If ASCII faster to use case insensitive regex than allocating using to_uppercase

Add ASCII fast path for ILIKE scalar (90% faster) #3306

Add ASCII fast path for ILIKE scalar (90% faster) #3306

Conversation

tustvold commented Dec 9, 2022 • edited Loading

Which issue does this PR close?

Rationale for this change

What changes are included in this PR?

Are there any user-facing changes?

tustvold commented Dec 9, 2022

alamb Dec 9, 2022

Choose a reason for hiding this comment

tustvold Dec 9, 2022

Choose a reason for hiding this comment

alamb left a comment

Choose a reason for hiding this comment

tustvold commented Dec 9, 2022

tustvold commented Dec 9, 2022

alamb commented Dec 9, 2022

tustvold commented Dec 9, 2022

tustvold commented Dec 9, 2022

tustvold commented Dec 9, 2022 • edited Loading

tustvold commented Dec 9, 2022

tustvold commented Dec 9, 2022

alamb Dec 9, 2022

Choose a reason for hiding this comment

tustvold Dec 9, 2022 • edited Loading

Choose a reason for hiding this comment

alamb left a comment

Choose a reason for hiding this comment

tustvold commented Dec 9, 2022 •

edited

Loading

tustvold commented Dec 9, 2022 •

edited

Loading

tustvold Dec 9, 2022 •

edited

Loading