Re: [HACKERS] build of 7.2.1 on SCO Openserver and Unixware

Peter Eisentraut <peter_e@gmx.net> writes:
> Do we need this at all?  Why not simply fix the expected files to remove
> the whitespace differences?

There shouldn't *be* any whitespace differences in a successful test.
The point of the -w switch is not to make any difference in a successful
test, it is to reduce the amount of irrelevant stuff printed in a failed
test.  If you have one bogus output item that is wider than the correct
value, that can cause psql to reformat the display wider in all rows.
The point of using -w is to not show those other rows as changed.

            regards, tom lane