MySQL 5.1.23 Regression Bugs

I've started testing the new Release Candidate (5.1.23) today. This was a version with a huge amount of fixes relative to latest MySQL releases - but I didn't expect to run into two bugs this fast.
  • Bug #34626 which I just filed in today and was already verified. This is a basic scenario which is used in the deployment scripts we use at Delver. Just try and pipe an empty file into the mysql command line client - and it crashes.
  • Bug #32633 which was apparently fixed in 5.1.24, and should have also presented itself in 5.1.22, but didn't for some reason. The upgrade brought it up.
    This one basically means that you just can't create stored procedures at all, unless you clean the sql_mode variable from certain tags. Which was really frustrating to find. I almost filed another bug but managed to find someone who already filed a duplicate 6 days ago.
    The SQL Mode is a session/global variable that determines which SQL compatibility mode the current session works as. More details here if it interests you.
I could say that I'm doing the community a favor by testing products or whatnot, but I'd just prefer the QA procedures at MySQL AB to be a bit stricter - I consider these as really trivial scenarios.
Conclusion: new versions fix old bugs and bring new ones. Trust no one. Test for yourself.

No comments:

Post a Comment