From: matz@... (Yukihiro Matsumoto)
Date: 2001-03-06T15:48:33+09:00
Subject: [ruby-talk:12096] Re: Another hook

Hi,

In message "[ruby-talk:12095] Re: Another hook"
    on 01/03/06, Dave Thomas <Dave@PragmaticProgrammer.com> writes:

|> And performance would be the excuse again.
|
|Would it be that bad in reality? You'd only have to do the call if
|to_bool was defined anywhere, so for programs that didn't use it it
|would only be a single instruction added to the path. I'm sure we can
|find one to take away somewhere to compensate... :)

If there's none, it must be OK.  But for example, if there's only one
to_bool among the required libraries, all conditional tests would be
slowed, unless I misunderstood something.

The boolean check is done by the RTEST() macro in the source now.  It
used to be a function (I mean a C function).  When I changed this into
bit-tweaking macro (in fact, Nobuyoshi Nakada sent me a patch),
performance was improved by several percent.  It depends on the view
whether it's enough price to pay, and on how much often to_bool used.

							matz.