Re: here's a good one from dizwell on the recent product launch

On Jul 19, 5:50 am, Mark Townsend <markbtowns..._at_sbcglobal.net> wrote:
> I looked at this one tonight. The problem boils down to the following
> reproducible case
>
> select 1 from dual
> where ('Tout' = CASE 'Oui'
> WHEN 'Oui' THEN 'Y'
> WHEN 'Non' THEN 'N'
> ELSE 'Tout' END)
> /
>
> Exactly as this - i.e it has to be literals used in this manner, with
> this outcome.
>
> This was reported in Feb 01, diagnosed by Feb 28, and assigned to a
> developer to fix. Based on perceived priority it was fixed and checked
> into the 11g code line in June. Priority was set low basically because
> the query is pretty darn stupid when you look at it. In this case, it
> was generated by an end-user access tool, which is presumably doing some
> sort of weird type of dynamic query macro processing to force a certain
> behavior that works across more than one database implemenation.

Yes it's a stupid query - though describing business objects as a weird end-user tool is perhaps a little harsh. Yes it does bad things, it's users do bad things - but they are still valid things and they tend to be for fairly valid business reasons.

In the end though I still stick with the fact that it was an example of a class of bug (wrong results) that go to the heart of why someone buys an rdbms product in the first place that doesn't get fixed rapidly - personally I find three classes of bug particularly pernicious and that sometimes, not always - appear to get treated lightly by Oracle. For the record they are

wrong results.
database irrecoverability - eg.Note 434527.1, or the whole class of rman errors with ntfs mounts that started with 10.2.0.x - nfs volume not mounted with correct options.
security fixes.

> However, I really don't think this bug is indicative of what Nuno is
> talking about.- Hide quoted text -

well having now seen the blog I agree - it looks like there's a whole raft of FBI and peoplesoft related issues at the heart of the problem. I'm not going to go into application patching since I've yet to see a vendor - including Oracle but also including their major competitors that makes application patch management as simple as it should be.

Niall Received on Thu Jul 19 2007 - 16:43:43 CDT