Subject Re: [IBO] Some further IBO/FB1 incompatibility considerations
( Try and resend the right one as it bounced )

> In first case the bug (ambiguous join) was fixed and actually every responsible programmer had to fix his code also, because it may have give wrong results sometimes.
> In second case there is no bug in applications and no need to fix.
> Very few people had ambiguous joins in their app, but nearly all apps that used IBO will be influenced by second case

I think that sums itr up.

> I know but I'm surprise for this situation:
> a) FB RC2, new syntax fix
> b) Marco Menardi (me) had to fix lot of queries with bad join syntax and test the apps (wery time consimung)
> c) none in the community was afraid about "old good apps" that can be broken by this fact
> ---

But - your applications would probably have failed safety
testing also because the results would have been
inconsistant. Difficult to test for, but something that I
have ahd to fix previously.

> d) Insert problem
> e) Marco Menardi (me) had to fix IBO code and recompile (5 minutes)
> f) lot of community members screaming loud about "broken compatibility" of old code...

Existing applications that fail now, may or may not be
broken, we do not know. It would be nice if we could
identify what is failing because of the join problem, and
what is simply being prevented from running but does not
have a 'real' problem

Marco - I can understand where you are coming from, and
simpathize. If the latest release of Firebird is NOT
ammended, then I will be stuck with 0.9.4 for some time to
come, and other users will be stuck with 5.6. The idea was
to get a stable build that we could move to without
problems. There will be some failures, but hopefully that is
because of real problems producing incorrect results, that
people may not even have known about. Something that causes
ALL programs to fail without a real error is just an added
complication we could do without.

Lester Caine
L.S.Caine Electronic Services