Hi Dean,

>That is sort of my point too. In my test, the longest time for a single line was 0.002. This was about 6 lines out of a couple of hundred. A particular method took as long as 0.4?? something, but its individual lines mostly show 0.000. Not much use for optimizing code!

Well, this tool is stored in a Coverage.App and the menu says "Coverage Profiler". Coverage is the basic job of this tool, profiling an additional feature. For technical reasons, the precision of the profiler is not very high, it's just like Seconds(), which under NT is only precise to a 1/100st second. That's by far not enough to profile a single line.

When I find a slow method but can't figure out which part of the method does make it slow, I go back to the old method. That is:

a) I add code to measure an entire codeblock:

   nStart = Seconds()
   * several lines of code
   nEnd = Seconds()
   DebugOut "Block #1"+Str(nEnd-nStart,10,3)

b) I measure single lines in a loop.

Christof