Re: Clean way to do string.join

Craig,
I think what is more interesting from an academic viewpoint is the use
of Format() vs. two Appends. The Append().Append() is twice as fast as
your original, and nearly twice as fast as the version that Stuart put
forward.

At the end of the day it really comes down to does it work, and is it
understandable? I doubt that the original poster needed anything other
than the solution that Craig put forward.

Cheers,
John


On Wed, 26 Jan 2005 14:48:00 -0800, Craig Wagner <craig.wagner@xxxxxxxxx> wrote:
>  > Sounds like a great interview question!
>  
>  Sounds like a great time-waster...
>  
>  Okay, I had some time to kill this afternoon so I put together a
>  simple benchmark.
>  
>  I iterated each option 1,000,000 times. I did this ten times and got
>  the average. Here's the results.
>  
>  ======================
>  StringBuilder result = new StringBuilder( 50 );
>  
>  for( int i = 0; i < value.Length; i++ )
>  {
>      result.Append( ( result.Length > 0 && value[i].Length > 0 ? " AND
>  " : "" ) + value[i] );
>  }
>  
>  return result.ToString();
>  
>  1093.5 milliseconds
>  ======================
>  string.Join( " AND ", CopyNonempty( value ) )
>  
>  1457.8 milliseconds
>  ======================
>  StringBuilder sb = new StringBuilder();
>  int i = 0;
>  for ( ; i < value.Length && value[i] == string.Empty; ++i ) ;
>  
>  if ( i < value.Length )
>  {
>      sb.Append( value[ i++ ] );
>      for ( ; i < value.Length; ++i )
>          if ( value[i] != string.Empty  )
>              sb.AppendFormat ( "{0}{1}", separator, value[i] );
>  }
>  
>  return sb.ToString();
>  
>  975.1 milliseconds
>  ======================
>  StringBuilder sb = new StringBuilder();
>  int i = 0;
>  for ( ; i < value.Length && value[i] == string.Empty; ++i ) ;
>  
>  if ( i < value.Length )
>  {
>      sb.Append( value[ i++ ] );
>      for ( ; i < value.Length; ++i )
>          if ( value[i] != string.Empty  )
>              sb.Append(separator).Append(value[i]);
>  }
>  
>  return sb.ToString();
>  
>  542 milliseconds
>  ======================
>  
>  So it would appear that Stuart's solution is indeed more efficient
>  than my original proposal, by the margin of 118.4 milliseconds OVER
>  ONE MILLION EXECUTIONS!  That works out to 0.0001184 milliseconds. You
>  can figure out the math for seconds.
>  
>  This time will add up over the next few hundred years.
>  
>  Granted, some applications will benefit from this type of
>  optimization, but my guess is that the vast majority could better
>  spend their time optimizing their database access or some other aspect
>  of the application.
>  
>  Sorry, but I worked with someone at my last company who would spend
>  hours, if not days, trying to optimize this kind of thing, totally
>  ignoring much larger performance issues in the application that would
>  get a much bigger bang for the buck.
>  
>  
>  ________________________________
>  Yahoo! Groups Links
>  
> To visit your group on the web, go to:
> http://groups.yahoo.com/group/padnug/
>   
> To unsubscribe from this group, send an email to:
> padnug-unsubscribe@xxxxxxxxxxxxxxx
>   
> Your use of Yahoo! Groups is subject to the Yahoo! Terms of Service.