Three Monkeys, Three Typewriters, Two Days

October 27, 2010

Nicholas Nethercote is a hero

Between LIR comments, fixing the flushing behavior of LIR and tracemonkey dumps, and annotating bytecode boundaries in the LIR, reading tracemonkey debug spew will become amazingly better. This will make it much easier to see which bytecodes get compiled in ways that are suboptimal

Now if only I could get per-instruction profiler blame for the generated code and then correlate it with these dumps.... Then we could start getting a pretty good idea of what things are hot and need work.

Posted by bzbarsky at 11:09 PM