that's not what "relational" says, it only strips whitespace from
values taken from the message, in other words the left hand side.
OK, let's think about this sensibly, though. I could ask Wolfgang, but
I'm quite sure that when he wrote the spec, he just assumed that no one
would write a test that could not possibly match, and if white space is
stripped from the data from the message, a string with leading white
space that's coded in the script can't possibly match it.
The other way around is a pretty stupid case too -- if we strip white
space from strings in the script and not from those from the message,
we'll have messages that can't possibly be matched by any script.
So, clearly, either we strip both, or neither.
Now that we have that settled, we're back to the real question, which
is whether to strip white space.
As to your example, well, I didn't ask whether you could *contrive* a
case where it could matter -- clearly, one can. I asked whether it was
a sensible thing to expect in reality, and I think it's not. And my
experience tells me that it's not. And my experience tells me that
allowing for it causes problems, over time.
If we don't, then someone's implementation is going to turn
To: kjetilho(_at_)ifi(_dot_)uio(_dot_)no
into " kjetilho(_at_)ifi(_dot_)uio(_dot_)no", and it'll be forever before
someone sorts
that bug out.
But if consensus has it that we don't strip, then, OK, we don't strip.
But then we "don't strip" on BOTH sides of the equation. Both, or
neither.
Barry
--
Barry Leiba, Pervasive Computing Technology
(leiba(_at_)watson(_dot_)ibm(_dot_)com)
http://www.research.ibm.com/people/l/leiba
http://www.research.ibm.com/spam