On 23/09/06, chromatic <[EMAIL PROTECTED]> wrote:
On Saturday 23 September 2006 11:30, A. Pagaltzis wrote:
> * Fergal Daly <[EMAIL PROTECTED]> [2006-09-23 19:35]:
> > At least then the user knows there's a problem _before_ he
> > <insert misfortune caused by non-functioning module>. Remember,
> > this thread is about how the toolchain is really for the user's
> > benefit. Hiding failures to avoid reports about known bugs is
> > for the developer's benefit, not the user,
>
> Maybe. Maybe not. When used to mean "todo" according to your
> interpretation, then yes. When used for one of various other
> use cases, then, just as likely, no. It depends on the author's
> intent in using PRETEND_OK – which is what I was saying. :-)
Agreed. I might add TODO/PRETEND tests but never document the feature they
exercise, so I haven't made any promise about what the module should do. I
don't think that warrants the harness saying "This module might not work as
you expect."
If you haven't documented it and nobody is using it yet then it's
perfectly reasonable to make that a TODO and in fact you have todo
items - make it work and document it - whether you have any intention
of ever doing them is another matter.
Where I have a problem is with the examples given in this thread which
have suggested using TODO to temporarily stop testing existing
functionality that users may already be depending on,
F
It might not, as people often expect crazy things, but I'm not sure that the
harness can or should assume why I added those tests.
-- c
