One thing I believe was true in the COBOL+JCL world of the 1980s is that you couldn't write a program that would read the name of a file, and then open that file. All files to be accessed had to be specified statically with JCL.
I was part of the Postgres Research Group at UC Berkeley from 1991 to ~1995, working directly for Mike Stonebraker. To be honest, I didn't do any of the research work behind it, although I did port Postgres to Windows NT during this time.
Postgres back then was completely different then what it is now. It was mainly used for PhD and Master's student to hack on for their research. It was a mess internally and was hardly usable for production. I don't think that MySQL went through this style of development.
What eventually happened is that around 1995 SQL was added to Postgres and a bunch of non-Berkeley people started hacking on it. They did a fantastic job, and deserve all the credit for making it what it is today (Stonebraker has publically said this).
Years ago I worked in the VAX/VMS development group at Sybase where SQLServer originated. SQLServer was basically an SQL interpreter consisting of several layers of loops. The innermost loop was written in assembler for speed.
I was able to remove one (1) machine language instruction from that innermost loop. I no longer recall if this resulted in a measurable difference but I've always been proud of this.
Am I remembering correctly?