On Mon, Nov 14, 2005 at 11:01:19AM -0500, Christopher Browne wrote:

> For instance, if you have a Really Long cluster namespace name,
> auto-created PK fields start breaking; I have added a check for 1.2
> which refuses to allow the Really Long cluster name.  Unfortunately,
> things we expect to break aren't nearly as easy to test :-(.

Why not?  If there's some code that checks for certain conditions,
you can check at the bounds of those conditions for the relevant
error, no?  I don't think it's reasonable to expect 100% coverage --
that's not what simple regression tests do -- but you can at least do
some minimal checks.

A

-- 
Andrew Sullivan  | ajs at crankycanuck.ca
"The year's penultimate month" is not in truth a good way of saying
November.
		--H.W. Fowler