Nearly one year ago, I found out that there is a bug in one of the core scripts which has to be executed after applying each Oracle 10g major patch. This bug was causing indefinite hang of the script, so the DBA was not able to upgrade the database on some platforms (for example x86 RHEL4).

The problem was really simple, component upgrades started with:

COLUMN patch_name NEW_VALUE patch_file NOPRINT;

And the script continued with lines like this one:

SELECT dbms_registry_sys.patch_script('JAVAVM') AS patch_name FROM DUAL;

The first SELECT call caused an indefinite hang. Simple solution to a simple problem was renaming the alias patch_name to eg. file_name in the whole script.

I did it, happily executed the script again and submitted the patch to Oracle Metalink. They haven’t paid for it yet and I was wondering, why they can’t pay users of their software for finding such a nasty bug. This should have been already done by internal testers many months before I found it. And testers were probably paid. So why can’t they also pay me?

Well, I accepted later that they can’t pay every user for finding a bug. Just because nearly no software is error free and Oracle RDBMS has the same error rate as any other. No, that’s not truth, there is one company producing even more buggy software and that company gave me the reason for using various Linuxes. But, that’s a different story.

What I also accepted was that sometimes it’s enough to say thanks, when user not only reports a bug, but also fixes it like I did. Of course the bugs have to be fixed by internal developers in such a respected company like Oracle, Inc. selling database servers for incredibly high prices worldwide. But you know, the developers are probably overloaded, underpaid, poorly managed and there can be of course much more excuses for not fixing each bug using their own brains. And in addition to that, I love to help global companies. Sometimes.

But there is still one piece of the puzzle which can’t fit in. I really can’t understand why that patch was not applied.

I had to run the upgrade once again on a different machine today. And you know what? Mentioned buggy script was changed five times by three different people since I found the bug and sumbitted the patch. It was also renamed from catpatch.sql to c1002000.sql. But it still contains the same bug. Nobody applied the patch. Even if I served it on a silver plate.

Now in my humble opinion, the time to pay has come. Not only for finding the but, reporting it and submitting a patch. But for publishing this article which will do the same service for the rest of the huge crowd of tearful database administrators who are using Oracle 10g just because there is no better database server out there. It’s sad, but true. Or not, Larry?