>>>>>That's fine, thanks again - I'll see your answer Saturday night. Let me know if you want to see the programs and log file.
>>>>>There are only 2 prg and the log.
>>>>
>>>>
>>>>This has gotten to a point where the only sensible answer I can give is
>>>>(1) take it to iteration 500
>>>>(2) Start coverage logging
>>>>(3) let it do the iteration
>>>>(4) stop coverage logging
>>>>
>>>>Then see which parts take a long time
>>>
>>>I've made a few changes in the write_log function and the time of execution is no longer increasing in my tests.
>>
>>
>>Good - exit memory leak
>
>If I can do better - I'd like to do better. I'll play with your fwrite idea.
Yes, - but another important point is appending each time to this.cInvokeFullLog. As it gets bigger it will slow down more
3 possibilities in order of speed
(1) get rid of cInvokeFullLog
(2) write it to another file with fcreate() and fputs()
(3) use an ADO stream
Gregory