How is performance of this compared to Go's regex?

[JensRantil]

Have you done any basic benchmarks?

[JensRantil]

Looking at https://godoc.org/github.com/dlclark/regexp2, I suspect this library is a lot slower since working with bytes tend to be a lot faster than strings.

[dlclark]

I have ported several benchmarks from the regexp package to regexp2 for comparison. You can find them in the regexp_performance_test.go file. Here's the output from my machine (2015 2.5Ghz i7 MacBook Pro)

BenchmarkLiteral-8                        	 3000000	       451 ns/op
BenchmarkNotLiteral-8                     	  300000	      4640 ns/op
BenchmarkMatchClass-8                     	 1000000	      1046 ns/op
BenchmarkMatchClass_InRange-8             	 1000000	      1036 ns/op
BenchmarkAnchoredLiteralShortNonMatch-8   	10000000	       158 ns/op
BenchmarkAnchoredLiteralLongNonMatch-8    	20000000	        98.5 ns/op
BenchmarkAnchoredShortMatch-8             	 5000000	       293 ns/op
BenchmarkAnchoredLongMatch-8              	10000000	       229 ns/op
BenchmarkOnePassShortA-8                  	 1000000	      1092 ns/op
BenchmarkNotOnePassShortA-8               	 1000000	      1130 ns/op
BenchmarkOnePassShortB-8                  	 2000000	       724 ns/op
BenchmarkNotOnePassShortB-8               	 2000000	       738 ns/op
BenchmarkOnePassLongPrefix-8              	 5000000	       318 ns/op
BenchmarkOnePassLongNotPrefix-8           	 5000000	       295 ns/op
BenchmarkMatchEasy0_32-8                  	20000000	        95.6 ns/op	 334.80 MB/s
BenchmarkMatchEasy0_1K-8                  	 5000000	       249 ns/op	4100.65 MB/s
BenchmarkMatchEasy0_32K-8                 	  200000	      7388 ns/op	4435.27 MB/s
BenchmarkMatchEasy0_1M-8                  	    5000	    278079 ns/op	3770.77 MB/s
BenchmarkMatchEasy0_32M-8                 	     100	  11788037 ns/op	2846.48 MB/s
BenchmarkMatchEasy1_32-8                  	10000000	       200 ns/op	 159.84 MB/s
BenchmarkMatchEasy1_1K-8                  	  300000	      5126 ns/op	 199.75 MB/s
BenchmarkMatchEasy1_32K-8                 	   10000	    146121 ns/op	 224.25 MB/s
BenchmarkMatchEasy1_1M-8                  	     300	   4805608 ns/op	 218.20 MB/s
BenchmarkMatchEasy1_32M-8                 	      10	 155318543 ns/op	 216.04 MB/s
BenchmarkMatchMedium_32-8                 	 3000000	       518 ns/op	  61.71 MB/s
BenchmarkMatchMedium_1K-8                 	  100000	     16009 ns/op	  63.96 MB/s
BenchmarkMatchMedium_32K-8                	    3000	    521474 ns/op	  62.84 MB/s
BenchmarkMatchMedium_1M-8                 	     100	  16629170 ns/op	  63.06 MB/s
BenchmarkMatchMedium_32M-8                	       2	 536617056 ns/op	  62.53 MB/s
BenchmarkMatchHard_32-8                   	  100000	     20548 ns/op	   1.56 MB/s
BenchmarkMatchHard_1K-8                   	    2000	   1010383 ns/op	   1.01 MB/s
BenchmarkMatchHard_32K-8                  	      50	  35005369 ns/op	   0.94 MB/s
BenchmarkMatchHard_1M-8                   	       1	1175492392 ns/op	   0.89 MB/s
BenchmarkMatchHard_32M-8                  	       1	38285374740 ns/op	   0.88 MB/s
BenchmarkMatchHard1_32-8                  	  500000	      2924 ns/op	  10.94 MB/s
BenchmarkMatchHard1_1K-8                  	   10000	    104998 ns/op	   9.75 MB/s
BenchmarkMatchHard1_32K-8                 	     500	   3484198 ns/op	   9.40 MB/s
BenchmarkMatchHard1_1M-8                  	      10	 110913710 ns/op	   9.45 MB/s
BenchmarkMatchHard1_32M-8                 	       1	3534549771 ns/op	   9.49 MB/s
BenchmarkLeading-8                        	  100000	     15429 ns/op
BenchmarkParserPrefixLongLen-8            	     100	  14529825 ns/op

And from the stdlib on the same box (go 1.12.1):

pkg: regexp
BenchmarkFind-8                           	10000000	       226 ns/op	       0 B/op	       0 allocs/op
BenchmarkFindAllNoMatches-8               	20000000	       121 ns/op	       0 B/op	       0 allocs/op
BenchmarkFindString-8                     	10000000	       222 ns/op	       0 B/op	       0 allocs/op
BenchmarkFindSubmatch-8                   	 5000000	       314 ns/op	      48 B/op	       1 allocs/op
BenchmarkFindStringSubmatch-8             	 5000000	       296 ns/op	      32 B/op	       1 allocs/op
BenchmarkLiteral-8                        	20000000	        78.3 ns/op
BenchmarkNotLiteral-8                     	 1000000	      1284 ns/op
BenchmarkMatchClass-8                     	 1000000	      1678 ns/op
BenchmarkMatchClass_InRange-8             	 1000000	      1646 ns/op
BenchmarkReplaceAll-8                     	 1000000	      1159 ns/op
BenchmarkAnchoredLiteralShortNonMatch-8   	20000000	        64.7 ns/op
BenchmarkAnchoredLiteralLongNonMatch-8    	20000000	        85.1 ns/op
BenchmarkAnchoredShortMatch-8             	20000000	       111 ns/op
BenchmarkAnchoredLongMatch-8              	10000000	       217 ns/op
BenchmarkOnePassShortA-8                  	 3000000	       509 ns/op
BenchmarkNotOnePassShortA-8               	 3000000	       464 ns/op
BenchmarkOnePassShortB-8                  	 5000000	       389 ns/op
BenchmarkNotOnePassShortB-8               	 5000000	       331 ns/op
BenchmarkOnePassLongPrefix-8              	20000000	        90.2 ns/op
BenchmarkOnePassLongNotPrefix-8           	 5000000	       302 ns/op
BenchmarkMatchParallelShared-8            	30000000	        59.1 ns/op
BenchmarkMatchParallelCopied-8            	20000000	        60.9 ns/op
BenchmarkQuoteMetaAll-8                   	20000000	        92.3 ns/op	 151.73 MB/s
BenchmarkQuoteMetaNone-8                  	30000000	        48.1 ns/op	 540.88 MB/s
BenchmarkMatch/Easy0/32-8                 	20000000	        58.0 ns/op	 551.51 MB/s
BenchmarkMatch/Easy0/1K-8                 	 5000000	       259 ns/op	3939.41 MB/s
BenchmarkMatch/Easy0/32K-8                	  300000	      4134 ns/op	7924.96 MB/s
BenchmarkMatch/Easy0/1M-8                 	   10000	    220826 ns/op	4748.42 MB/s
BenchmarkMatch/Easy0/32M-8                	     200	   7406666 ns/op	4530.30 MB/s
BenchmarkMatch/Easy0i/32-8                	 2000000	       953 ns/op	  33.58 MB/s
BenchmarkMatch/Easy0i/1K-8                	   50000	     25836 ns/op	  39.63 MB/s
BenchmarkMatch/Easy0i/32K-8               	    2000	   1072221 ns/op	  30.56 MB/s
BenchmarkMatch/Easy0i/1M-8                	      50	  34027283 ns/op	  30.82 MB/s
BenchmarkMatch/Easy0i/32M-8               	       1	1147085221 ns/op	  29.25 MB/s
BenchmarkMatch/Easy1/32-8                 	30000000	        54.7 ns/op	 585.51 MB/s
BenchmarkMatch/Easy1/1K-8                 	 2000000	       718 ns/op	1424.34 MB/s
BenchmarkMatch/Easy1/32K-8                	   50000	     30969 ns/op	1058.09 MB/s
BenchmarkMatch/Easy1/1M-8                 	    2000	   1085243 ns/op	 966.21 MB/s
BenchmarkMatch/Easy1/32M-8                	      50	  34915840 ns/op	 961.01 MB/s
BenchmarkMatch/Medium/32-8                	 2000000	       820 ns/op	  39.01 MB/s
BenchmarkMatch/Medium/1K-8                	   50000	     24286 ns/op	  42.16 MB/s
BenchmarkMatch/Medium/32K-8               	    2000	   1045737 ns/op	  31.33 MB/s
BenchmarkMatch/Medium/1M-8                	      50	  33503487 ns/op	  31.30 MB/s
BenchmarkMatch/Medium/32M-8               	       1	1075915924 ns/op	  31.19 MB/s
BenchmarkMatch/Hard/32-8                  	 1000000	      1251 ns/op	  25.56 MB/s
BenchmarkMatch/Hard/1K-8                  	   50000	     37643 ns/op	  27.20 MB/s
BenchmarkMatch/Hard/32K-8                 	    1000	   1474564 ns/op	  22.22 MB/s
BenchmarkMatch/Hard/1M-8                  	      30	  48328999 ns/op	  21.70 MB/s
BenchmarkMatch/Hard/32M-8                 	       1	1677001522 ns/op	  20.01 MB/s
BenchmarkMatch/Hard1/32-8                 	  200000	      7289 ns/op	   4.39 MB/s
BenchmarkMatch/Hard1/1K-8                 	   10000	    217006 ns/op	   4.72 MB/s
BenchmarkMatch/Hard1/32K-8                	     200	   6988508 ns/op	   4.69 MB/s
BenchmarkMatch/Hard1/1M-8                 	       5	 227978092 ns/op	   4.60 MB/s
BenchmarkMatch/Hard1/32M-8                	       1	7342015123 ns/op	   4.57 MB/s
BenchmarkMatch_onepass_regex/32-8         	 2000000	       707 ns/op	  45.24 MB/s	       0 B/op	       0 allocs/op
BenchmarkMatch_onepass_regex/1K-8         	  100000	     20132 ns/op	  50.86 MB/s	       0 B/op	       0 allocs/op
BenchmarkMatch_onepass_regex/32K-8        	    2000	    642681 ns/op	  50.99 MB/s	       0 B/op	       0 allocs/op
BenchmarkMatch_onepass_regex/1M-8         	     100	  20804244 ns/op	  50.40 MB/s	      11 B/op	       0 allocs/op
BenchmarkMatch_onepass_regex/32M-8        	       2	 660706838 ns/op	  50.79 MB/s	     564 B/op	       1 allocs/op
BenchmarkCompileOnepass-8                 	  300000	      4169 ns/op	    4016 B/op	      52 allocs/op

Usually the stdlib is faster, but not always. If you need the specific backtracking features I have found the performance is acceptable. If you don't use the backtracking features then you should just use regexp from the stdlib.