@stchang has joined the channel
reader in C: about 5k lines; reader in Racket: about 3.2k lines
regexp in C: about 6k lines; regexp in Racket: about 7k lines
:shrug:
regexp benchmarks
In case anyone is interested in the performance; bottom line: x1-x2 for typical uses, and x2-x12 for stress cases (and the benchmarks are mostly stress cases)
The upper end of the slowdown is always some iteration on a long byte string, such as looking for one particular character in a very long byte string
@samth I wonder how well Pycket does on the regexp matcher; it’s a kind of interpreter, but internally using a closure compiler, and my guess is that Pycket does well with that (as opposed to an interpreter dispatch)
I’ll try it out
Or really, I just invited @sabauma to this channel to do so
@sabauma has joined the channel
@samth The benchmark is this rx.zip file?
Yes
I haven’t looked at it yet, it’s late here
Those are all html files though.
Is the code to run in the linklets branch of Racket?
@mflatt: Where can I find the code used to generate those timings?
but I hacked it to add racket-impl….
Where can I find this file: "/Users/mflatt/linklet/pkgs/regexp/main.rkt"?