Hi Jonathan,
Silly question time - I assume that if you don't includes the DBI->connect line, then the two invocations of showbin produce the same output in both versions of Perl.
It does. And in that case, the only float representation obtained is: 0011111111111011110010101001011010010001101001110101110011010001
A wise programmer once said 'moving floating point numbers is like moving sand piles; every time you do, you lose a little sand and you pick up a little dirt'.
Sure. And others have made this point on this list and it is true. But in that particular case we are doing the *same operation* before and after, on the same start data. I just want to emphasize that the issue discussed in this thread is not related to floating point precision, at least not in the meaning of the usual loss of precisions following conversion to floating point and arithmetic operations on them. This issue is about how perl alters its behavior when converting a string to a native float, after some oracle modules have been loaded. See Tim's latest email for hints of the details behind this conversion.
