Well here's my thinking (which is probably a bit twisted...) on it and why this didn't occur to me at first...

Sure, I know it *returns* the same value, but that...

Yeah, that did it. Odd, though. Seems to me that the function should anticipate something like that.

Just to update, applying CU2 did indeed fix the issue.

As said, it's quite painful. You'd be better off to do a re-install if possible.

I think I found it:

http://support.microsoft.com/kb/959028

If I issue SET XACT_ABORT OFF; before hand the parse runs without error (for that session).

Dev machine is on 9.00.4035 so it makes sense I guess....

Same error (as above) for all three.

Everything else appears to run fine, it's just the "Parse" functionality.

I log in with sa so authentication issues are doubtful.

No relevant messages in the SQL logs or Windows event logs.

DBCC CHECKDB returns no errors for master, msdb or a few...