Jeffrey,

Can you accurately control the execution time per row? Can you store the average time it last took and use that to estimate?

For shorter processes I use the thermometer ActiveX.

When I have a long running process (more than a couple of minutes) I use a little EstimatedTimeLeft object that displays the clock time when the process will finish. The class logs the start time and number of total rows to be processed, then every row calls the Update() method. this calculates average time per row based on total elapsed time / # rows processed. Then every nTH iteration it updates the finish time display by mulitplying # of rows left to process * average time per row. I find this works well because someone can look at the display and it tells them "I'll be finished at 14:35" and the user can schedule the rest of their work.

This works well for processes that are linear in peformance. If the process starts out with say 10 seconds per row and then finishes at a rate of 1 second per row the counter will vastly over estimate the finish time. Conversely if the first rows go at 1 row per second and the last rows are 10 seconds per row it'll vastly under estimate the finish time.

>Does anyone know how i can get a count of records and give the users an estimated process time so that they can determine whether or not to proceed with a module?