On Mon, Mar 21, 2005 at 03:02:57AM -0500, Tom Lane wrote:
>
> One thing that needs some thought is how you are going to test this
> robustly. I'd not feel any great deal of confidence in a test that
> expects that we can push \r\n sequences into CVS and expect them to
> survive unmodified into distributed filesets. Every so often someone
> commits a DOS-newlines file into CVS, and they hear about it PDQ...
We're testing what happens when \r gets into prosrc, right? Shouldn't
creating the function with \r in single quotes instead of dollar
quotes work, as in the tests I showed in my original submission?
http://archives.postgresql.org/pgsql-patches/2005-03/msg00186.php
The functions with \r failed as expected in an unpatched system,
and they worked in a patched system. The CREATE FUNCTION statements
used \r escape sequences, so that shouldn't be a problem with CVS
since they're not actual carriage returns in the SQL, right?
--
Michael Fuhr
http://www.fuhr.org/~mfuhr/