aantron / markup.ml / 123

78%

push

travis-ci

Speed: buffer channel reads

The speedup is insensitive to the buffer size once it is over 64 bytes.
Presumably, that is the point at which the time to read the bytes
dominates the overhead of making the read call and handling its result.
However, I arbitrarily chose 4096-byte buffers.

Time to run this code

  Markup.(file "foo.ml" |> fst |> drain)

decreased by about 30%. Testing was done with Flambda -O3.

Related #15.

19 of 19 new or added lines in 1 file covered. (100.0%)

3327 of 4263 relevant lines covered (78.04%)

349.17 hits per line