From: William James <w_a_x_man@...>
Date: 2008-03-12T03:57:59+09:00
Subject: Re: duplicating characters in a string

On Mar 11, 9:26 am, Robert Klemme <shortcut...@googlemail.com> wrote:
> 2008/3/11, William James <w_a_x_...@yahoo.com>:
>
> > On Mar 7, 11:45 pm, Joel VanderWerf <vj...@path.berkeley.edu> wrote:
>
> >  > Rehearsal ---------------------------------------------------
> >  > s.gsub(re){...}   6.220000   0.010000   6.230000 (  6.325230)
> >  > s.gsub(re,...)    3.020000   0.060000   3.080000 (  3.115286)
> >  > 7stud             4.110000   0.000000   4.110000 (  4.142912)
> >  > 7stud             2.050000   0.000000   2.050000 (  2.072398)
> >  > ----------------------------------------- total: 15.470000sec
>
> >  >                        user     system      total        real
> >  > s.gsub(re){...}   5.910000   0.020000   5.930000 (  5.993456)
> >  > s.gsub(re,...)    3.000000   0.000000   3.000000 (  3.017119)
> >  > 7stud             4.100000   0.020000   4.120000 (  4.300772)
> >  > 7stud             2.030000   0.010000   2.040000 (  2.068190)
>
> > Pascal (FreePascal):
>
> >  uses sysutils;  // for timing
>
> >  function dup_chars( s: ansistring ): ansistring;
> >  var
> >   out: ansistring;
> >   i,j: longint;
> >   c: char;
> >  begin
> >   setlength( out, length(s) * 2 );
> >   for i := 1 to length(s) do
> >   begin
> >     c := s[i];   j := (i shl 1);
> >     out[j-1] := c;
> >     out[j] := c;
> >   end;
> >   exit( out )
> >  end;
>
> >  var
> >   s : ansistring;
> >   i : longint;
> >   when : tDateTime;
> >  begin
> >   s := '';
> >   for i:=1 to 1000000 do  s += 'abc';
> >   when := Time;
> >   dup_chars( s );
> >   writeln( ((time - when) * secsPerDay):0:3, ' seconds' )
> >  end.
>
> >   --->0.031 seconds
>
> Your point being?
>
> robert
>
> --
> use.inject do |as, often| as.you_can - without end

On Mar 11, 9:26 am, Robert Klemme <shortcut...@googlemail.com> wrote:
> 2008/3/11, William James <w_a_x_...@yahoo.com>:
>
> > On Mar 7, 11:45 pm, Joel VanderWerf <vj...@path.berkeley.edu> wrote:
>
> >  > Rehearsal ---------------------------------------------------
> >  > s.gsub(re){...}   6.220000   0.010000   6.230000 (  6.325230)
> >  > s.gsub(re,...)    3.020000   0.060000   3.080000 (  3.115286)
> >  > 7stud             4.110000   0.000000   4.110000 (  4.142912)
> >  > 7stud             2.050000   0.000000   2.050000 (  2.072398)
> >  > ----------------------------------------- total: 15.470000sec
>
> >  >                        user     system      total        real
> >  > s.gsub(re){...}   5.910000   0.020000   5.930000 (  5.993456)
> >  > s.gsub(re,...)    3.000000   0.000000   3.000000 (  3.017119)
> >  > 7stud             4.100000   0.020000   4.120000 (  4.300772)
> >  > 7stud             2.030000   0.010000   2.040000 (  2.068190)
>
> > Pascal (FreePascal):
>
> >  uses sysutils;  // for timing
>
> >  function dup_chars( s: ansistring ): ansistring;
> >  var
> >   out: ansistring;
> >   i,j: longint;
> >   c: char;
> >  begin
> >   setlength( out, length(s) * 2 );
> >   for i := 1 to length(s) do
> >   begin
> >     c := s[i];   j := (i shl 1);
> >     out[j-1] := c;
> >     out[j] := c;
> >   end;
> >   exit( out )
> >  end;
>
> >  var
> >   s : ansistring;
> >   i : longint;
> >   when : tDateTime;
> >  begin
> >   s := '';
> >   for i:=1 to 1000000 do  s += 'abc';
> >   when := Time;
> >   dup_chars( s );
> >   writeln( ((time - when) * secsPerDay):0:3, ' seconds' )
> >  end.
>
> >   --->0.031 seconds
>
> Your point being?
>
> robert
>
> --
> use.inject do |as, often| as.you_can - without end

On my computer:

s.gsub(re, ...)  --> 9.641000 seconds
Lua              --> 1.078   seconds

Ruby's gsub is far too slow.  7stud even beat it with an each_byte
loop!
Disgraceful.  Disgusting.  Inexcusable.

The Lua code:

s = string.rep( 'abc', 1000000 )

time = os.clock()
string.gsub( s, '(.)', "%1%1" )
print( os.clock() - time, "seconds" )