----- Original Message -----
From: "Conor MacNeill" <[EMAIL PROTECTED]>
To: "Ant Developers List" <[EMAIL PROTECTED]>
Sent: Monday, April 15, 2002 12:04 AM
Subject: RE: new bug w/ xmlvalidate


> > From: Steve Loughran [mailto:[EMAIL PROTECTED]
> >
> > We could add more failing tests in there, the aim being to have all
those
> > tests passing by ant 1.5 ship date.
> >
> > Seem like a good tactic?
> >
>
> +1
>
> Indeed, I like the approach of adding the testcase in first, then fixing
the
> code. I did this recently with a <depend> bug but held off checking in any
> of it until the bug was fixed. This approach will allow these two roles to
> be independant which would be good.
>
> The only caution would be to ensure it doesn't become a hiding place for
> "broken windows". Probably not much of a concern with the level of
scrutiny
> Ant gets.

The biggest problem I have encountered with having a set of tests that dont
work is that if they suddenly fail differently then you dont notice, because
you ignore them. But that does not mean that you should not have such tests,
only that you work to fix them, and run them regularly to make sure they are
still broken.

-steve


--
To unsubscribe, e-mail:   <mailto:[EMAIL PROTECTED]>
For additional commands, e-mail: <mailto:[EMAIL PROTECTED]>

Reply via email to