Making The Existing Ruby Interpreter Faster
A lot of people seem to want to reinvent the wheel where Ruby is concerned, and I wish them all luck, but Tomasz Węgrzanowski has taken the unique step of trying to make the existing C-based Ruby interpreter faster instead, with intriguing results.. He even goes as far as to hack Ruby's source code to remove inefficiencies in how Fixnum objects are compared. I totally dig this and would love to support further efforts to implement non-damaging optimizations to the existing interpreter.
Could someone intimately familiar with the construction of the current interpreter comment on whether these techniques en masse could yield significant benefits at little re-implementation cost?