>>>>> "B" == Brian Candler <B.Candler / pobox.com> writes:


B> That trivial test program was about twice as fast under Perl. I would guess
B> that calls to the I/O library represent only a small part of that difference

 See [ruby-talk:68578]

ruby168 BIG    = 319
ruby168 SMALL  = 95

ruby18p2 BIG   = 12
ruby18p2 SMALL = 12

B> - I don't think any changes should be made (particularly ones which might
B> sacrifice portability or make the code harder to maintain) without real,
B> hard profiling evidence.

 The code is not harder to maintain.


Guy Decoux