Understanding parsing performance

[cdietrich]

Hi, is there a way to understand where the parser actually spends time
looking at a cpu prof in speedscope we cannot really see if there is any of our rules we might be able to optimize

[msujew]

I believe we should cover this under #1124. We need better logging anyway.

[cdietrich]

@msujew i did some profiling by manually patching subrule and recording types on rule calls
there i can see that alternatives seems to be quite expensive.
is there some stragegies to optimize it?

[ydaveluy]

Hi,

I think that a profiler that print how many time each rules is called with duration would be very usefull.

FYI in this parser https://github.com/yhirose/cpp-peglib there is a profiling mechanism (available in playground).