----- Original Message ----- From: Michael G Schwern <[EMAIL PROTECTED]>
> Finally, this argument seems to be that "not ok # TODO" > should fail which is to say "don't use TODO tests". I reread what I wrote and I left an extra sentence in there which I didn't mean, hence the confusion. 'not ok # TODO' should pass. If 'ok # TODO' failed and the author was using a sane tool, the author would see the broken test before shipping. > In order for this scenario to happen the *developer*, > not the user, is the one making the mistakes. Agreed. Developers do that all the time. > The developer must release code with a stub in it. Nope, never happened in the history of programming. > The developer is the one who is using the test > harness which does not display TODO tests. > The developer shipped busted code. I wasn't meaning to suggest 'do not display TODO tests'. I'm suggesting 'correct broken behavior'. > At that point I'd say "your shit is broke, use a better harness". No comment :) > And even with your proposed change, where passing > TODO tests become failures, the developer could > misinterpret it just as your proposed user does. On this point we agree. However, I'd suggest that the developer is more likely to know his or her code than an end user. Cheers, Ovid
