> This article is in desperate need of revision. Its message would be
> greatly improved if the parts about readline were removed and you focused
> solely on the OpenSSL issue.
Well, I'm sorry it's confusing, that is (obviously) not the intent. But the problem involves the combination of PostgreSQL (or really *any* program), OpenSSL *and* readline. With only one of the latter two (OpenSSL or readline) being linked into PostgreSQL, there would be no problem.
PostgreSQL was just the most recent place where the issue has come up, but linking OpenSSL and GPL code has been a recurring problem (in the eyes of some, anyway). In this case, the GPL code is readline.