On Fri, Mar 21, 2014 at 10:39 AM, Silvan Jegen <s.je...@gmail.com> wrote: > "Benchmarking" the decoder with > > time for i in `seq 10000`; do cat UTF-8-test.txt; done;
Turns out that was a stupid idea since this approach probably just measures the time cat needs to start x10000. If I create a file containing the content of UTF-8-test.txt 10'000 times (resulting in a 2'981'000 line file) and cat that one the old decoder uses around 17-18 second while the new one takes around 18-19 (so the new one may be slightly slower). I ran the command only around 5 times so if the suckless community thinks performance is critical in this case we would need to do some proper benchmarking.
