Monday, August 22, 2005

Having ASORT of Frustrating Day

Have you ever had one of those days where nothing seems to go your way? Today was one of those days for me. I hit three deadends on three different projects today. We are definitely near a full moon. When this happens I put the project down for a bit and hope for inspiration or divine intervention, and most of the time it works.

About a year ago a developer who uses ViewEditor sent me a bug report noting the list of tables and views in the Copy Columns to Clipboard feature was not sorted properly. I looked into it and could not reproduce the problem. He was using the VFP 9 beta at the time and we both passed it off as a possible bug in the beta. Frustrating for both of us, but I kept the report around in case I got another report or better steps to reproduce the problem. It is not a show stopper type bug, but definitely causes frustration when working with lots of views or lots of tables in a DBC.

Fast forward one year. No one else reported this problem, well, until a week or so ago. The second developer is using the production version of VFP 9. The results were the same on my end as I could not reproduce the problem based on the report. One report could be a fluke, but two is doubtful. So I spent 3 hours while I was on my way home from vacation trying to figure out some reproducible steps to show Microsoft how ASORT() is broken. I could not get it to fail. Searching the Microsoft Knowledgebase and a post on Tek-Tips turned up nothing.

This morning I wrote a quick program for the developer to run which created the array using ADBOBJECTS() and sorted it using a couple of different collation sequences. Sure enough the results on the other developer's machine were identical to mine. This drove another couple of questions, which triggered additional information, which lead me to find the bug in the ViewEditor. Finally. Once I had the complete information I was able to develop the test condition and plain as day, there was a list of tables out of order. Took me less than 15 minutes to find the bug, make the fix, and test it.

The good news is the bug is finally found. The better news is I will be starting a beta for ViewEditor v3.6 soon with the highly respected White Light Crash Test Dummies. I have nearly a dozen enhancements and a half dozen bug fixes almost ready to roll.

0 Comments:

Post a Comment

<< Home