When dBase IV came out (From Ashton Tate, Not Borland) it was buggy, and much of the stuff in the user manual literally did not work. I was super excited about the SQL support in dBase IV - until after an entire evening of trying to get it to work I realized they had shipped non-functional code. Not buggy - it just didn't actually work as documented. Ended the company. (They came up with massive patches years later - but it was too late - people had moved on).
I would love to read a story on how they decided to ship a product that was essentially still 12-18 months from being completed. I'm guessing (?) they were about to run out of cash - and selling something, even if it didn't work, kept the lights on? Hard to believe given how dominant dBase III+ was back then.
[1] Where, by ton, I mean many, many hundreds of dollars. Which felt like a lot of money back then.