--- In firebird-support@yahoogroups.com, Jason Dodson <jasond@...> wrote:

>
> > It is amazing just how much 'funny' data gets into a live

database, even

> > with input checking and the database checks themselves. Personally I
> > will not be surprised if you find something in the data that is

'adding

> > to the test case' which we don't know about yet :)
> >
>
> Which I assume is the case.
>
> And that is what I was trying to get at Adam (If you are reading).

It wasn't that I didn't appreciate your efforts, it

> is just that I don't think going for the throat on this one is how

we will figure this out.

>
> Thanks,
> Jason

Jason,

Let me make something clear.

I hope you are able to reproduce the bug and I hope it does not take
too much effort. I want the bug to either get fixed or if it can not
be fixed in a timely manner to at least have the cause identified as a
'known issue' so I can avoid encountering the same problem.

You suggested that Firebird had a bug which meant that dependencies
were not enforced when changing data types on a field. I produced a
simple case which worked correctly.

You have to understand that not enforcing dependencies is a pretty
serious accusation against an ACID complient database engine (the
whole point of 'C'), and I imagine some of the developers may have
been a bit worried. I was worried myself, what if one of my patches
corrupts a database?

But this does not mean we give up and go home, it just means that our
hypothesis was incomplete and needs to be tweaked a little. Very
simple scientific method at play.

My comments were meant as an invitation for more information, and I am
sorry if they did not come across that way. The more information we
have, the more likely our hypothesis will be right. The more bugs we
can duplicate, the better Firebird can be as a result.

Adam