An analyst scheduled a meeting to talk about a problem discovered in production this year. They changed the format of some of the root level flags stored in the database. However, the fields that depend on those flags was still getting set using the old logic. Those fields in turn drive the setting of other fields that the customer uses in their reports.
The game plan was to update the function that sets the root level flags. We also will need to correct al the data previously processed incorrectly. Easy enough. A new guy was assigned to work on the root level flags issue. And I was tasked with updating the customer facing data.
The only tricky part of my fix was that it depended on the other fix being executed first. I asked the new guy to let me know when his fix was done so I could add a dependency to it. The new guy did contact me. He asked if I knew how to do a data update. I gave him an example script. It is a template I use when we update data in tables using some other tables as source.
The new guy too ka shot at his task. He asked me to review his work. Initially I saw some SQL errors in the code. He obviously did not run this code yet. I let him know about those problems. Then I noticed something weird. He was updating data in a table based on other fields in the same table. That did not seem like how we normally do things.
Upon closer inspection, I found that the dude did not understand which tables all the columns were in. I asked him to take a step back and look at the problem from a big picture perspective. I pointed out that normally data flows from table to table, not from a table to itself. You should understand what it is you are doing before trying to construct a technical solution to a problem.
