Going back to my last job that was mostly SQL, I was working in Perl, with SQL embedded in TT templates. Good luck finding an IDE that understands that.

In that job the comma placement was critical. Which column would be the last column would depend on what options the report was being generated with. The point is that people would ask for many variations on the same report. One person would want a top level view, another would want to break out event type, another would want to break it out by which type of advertising campaign was thrown at it - and all were being generated out of the same templated SQL so I didn't have to maintain code in parallel. But the result is that in the code I couldn't make assumptions about which column was going to wind up being last.

Thanks to leading comma formatting, it was sufficient to require that the FIRST field stayed the same. Which proved to never be a problem.