The problem with std.@

[ConorOBrien-Foxx]

After trying to identify why the empty program takes ~0.7s to run on my computer, I identified that the initialization of a state instance is the problem:

    def initialize(program, input=STDIN, output=STDOUT)
        @trees = ast(program)
        if @trees.nil?
            exit
        end
        @variables = @@default_variables.dup
        @abstract_references = []
        @locals = [{}]
        @saved = []
        @position = nil
        @in = input
        @out = output
        load_lib "std"
    end

Specifically, the last line load_lib "std". This loads the almost 800 line file std.@, which has a lot of useful functions in it. Most of the file is composed of comments.

I don't know what the particular problem is. Is it tokenizing? Is it node traversal?

This issue serves as a note for myself.

A potential solution is to design a way to compile Attache files....

[somebody1234]

Compilation is definitely a good idea. re: the performance problem, what does a profiler say

[ConorOBrien-Foxx]

@somebody1234 I haven't done any formal profiling. any suggestions?

[somebody1234]

Apparently the normal way to do it is ruby-prof (taken directly from repo):

RubyProf.start
# ... code to profile ...
result = RubyProf.stop

# print a flat profile to text
printer = RubyProf::FlatPrinter.new(result)
printer.print(STDOUT)

[somebody1234]

Reduced to ~0.25 seconds, can't reduce any further without significant changes to the lexer (which is still the bottleneck)