The idea here is that because a ParserSet is short-circuiting, that it will stop iterating over parsers once it's found one that works, that we can optimize for the common case by putting the common cases before the uncommon cases in the search order.
We could do this manually, but that requires periodically validating that our order still matches reality. Instead this pull request will have the ParserSet periodically optimize itself by recording how many
times each case was successfully used, and reorder the cases by how many times each one has been used.
The idea here is that because a
ParserSetis short-circuiting, that it will stop iterating over parsers once it's found one that works, that we can optimize for the common case by putting the common cases before the uncommon cases in the search order.We could do this manually, but that requires periodically validating that our order still matches reality. Instead this pull request will have the
ParserSetperiodically optimize itself by recording how many times each case was successfully used, and reorder the cases by how many times each one has been used.