From: Jean-Hugues ROBERT <jean_hugues_robert@...>
Date: 2002-04-12T15:40:18+09:00
Subject: Re: How to Make a Method Ineffective Efficiently?

Hello,

"an efficient debugging log facility"

In debug( msg) you don't want msg to be evaluated when debugging is off.

You may use a block that will get called only if debugging and will return 
the string to print: debug{ msg}. Note the {} instead of ().

Example:

def debug( &proc )
   if @@debug_debugging then
     @@debug_file.write "Debug: @{proc.call()}
     end
   end
end

then use it:

   debug{ "An inefficient msg: @{xxx()}

The idea is that the &proc gets computed *only* when in debugging mode.

You may get some extra speed by redefining debug() using eval() whenever 
the debugging flag changes. i.e. "def debug() end" when debugging is off.

Jean-Hugues

At 05:36 12/04/2002 +0900, you wrote:
>Hi,
>
>I am having a problem regarding coding efficiency (both for the execution
>time and for the typing itself).  Basically, I have a method that may or
>may not print depending on a flag, such as
>
>     ....
>     def print_info (*text)
>       if @debugging
>         ....
>         @file.write (text.join)
>       end
>     end
>
>Originally, I have put the flag in the calling class itself, but it made
>me typing the flag in too many places:
>
>     ....
>     print_info ("#{obj1} #{obj2} .... #{obj1000}") if @debugging #format A
>
>(because in reality the flag or condition may not be that
>short).  Therefore, just like in the first fragment of codes, I have put
>the flag in the print_info method itself, so that I can simply write
>
>     ....
>     print_info ("#{obj1} #{obj2} .... #{obj1000}") #format B
>
>and whether it actually prints or not is controlled by the flag in the
>print_info, which I can control without having to type in too many places.
>
>However, now I have a problem with execution efficiency.  With the last
>format (format B), whether or not I actually print, the methods to_s of
>obj1, ..., obj1000 are called.  The methods to_s themselves may be very
>complicated.  In the first format (format A), the methods to_s are not
>called at all if @debugging is equal to false.
>
>I don't want to type "if @debugging" everywhere (they clutter the codes),
>but I also want to have efficient execution, which in this case is
>equivalent to if @debugging is false, then the print_info method is
>ineffective, not even the passed arguments.  Is there any way I can
>accomplish what I want easily?
>
>(My current solution is to use format B, and try to minimize the argument
>overhead, by minimizing the number of arguments and doing processing
>inside the print_info method below the "if @debugging" line as much as
>possible.)
>
>Regards,
>
>Bill

---------------------------------------------------------------------------
Web:  http://hdl.handle.net/1030.37/1.1
Phone: +33 (0) 4 92 27 74 17