Hi,

>>
>> It's better than IB 6.5 -- IB 6.5 (supposedly tested for a couple of
>> days/longer? in the test suites) has the CHECK CONSTRAINT fires before
>> BEFORE INSERT trigger bug! Isn't that terrible!
>>
>
>IMO you can run IB against the test suite for weeks and it wont make the
>slightest difference. The test suite is flawed for many reasons. For
>one, it only tests what it knows about. If there is no test to verify
>the order of constraint/trigger firing then there is no way the suite
>can 'fail'.

But still, for example Craig Stunz is boosting about this being the
major difference between the open source version of IB or Firebird
and the certified version.

btw, this case _can_ be tested by simply creating a table and a
check constraint and a trigger, and it can be added to the basic
test.

>More seriously the fundamental principle of the test suite is based on
>comparing output from a test with stored output from a previous test
>run. There is no way of validating whether the test itself is valid, or
>whether the output itself is valid, beyond a QA engineer's say so.

A previous test result should be validated by a human begin, as so
with every automated test. And I am the last person to say that
everything can be automated and tested without human intervention.

>The test suite is not entirely meaningless - it can catch things where a
>new build may have regressed existing functionality - but overall it is
>nothing when compared with the twenty thousand downloads of Firebird 1.0
>RC2. The open source development model allows Firebird to be tested in
>constant daily use world-wide and for reasonably rapid fixes to be
>delivered when serious bugs are found.
>

I know that and you know that :) ... seems that not everyone at
TeamB (wants to) knows that :)


Looking forward to FB1 - thanks.

Martijn Tonies
InterBase Workbench - the developer tool for InterBase and Firebird
http://www.interbaseworkbench.com


[Non-text portions of this message have been removed]