|
Am 28.03.2013 14:37 schröbte mlepage@antimeta.com:
Wow thanks for that awesome work. I'll have to digest it and see which approach is best for my particular case. But having the approaches, and micro benchmarks, on the list is a great resource.
I was interested how good/bad a naive strcmp would perform. Btw. I didn't send the actual C code to the list because it is long, repetitive, and boring. If you want it for your own measurements, I can send it to you off-list.
Another interesting data point: sample1.so: 8.2k sample2.so: 8.2k sample3.so: 13k sample4.so: 8.2k sample5.so: 13k sample6.so: 8.2k So ragel and gperf increase the code size quite a bit ... Philipp