From: Phlip <phlipcpp@...>
Date: 2003-06-09T22:40:02+09:00
Subject: Re: Problem with "require" stmt in "test-first " tutorial

Dmitry V. Sabanin wrote:

> Richard wrote:

> > Question:  This output is a distortion from the output obtained when
> > running the script in a Command window.  It looks like SciTe treats
> > the output from the script proper via "puts" distinctly from output
> > from the "unit" package, so that they are presented out of order in
> > the output pane.
>
> I'm just a testing newbie, but i think that no output should be produced
from
> _automatic_ tests. That's not XP way i guess. All tests must be done via
> assertions.

The "XP way" does not specify the innards of the *Unit lineage of test rigs.

For historical (or possibly hysterical) reasons, those test rigs bunch up
the diagnoses from the assertions, and reveal them at the end.

I personally feel this is Big Test Up Front territory, and against the
ideals of TDD.

Specifically, TDD is completely interactive. The test rig is a programmer's
interface.

I may want to say assert(x); exit(0), run the tests only to that point, but
still read the diagnostics.

> > I'm happy to have Unit Tesing working, but I'd like to know the
> > details about SciTe so that I interpret output from future usage
> > correctly.

Done right ( ;-) there's no reason to say assert(x); exit(0). One should
only have a few diagnostics at a time, related to ones most recent edit.

Because TDD so often leads to doing things right, this hysterical feature
isn't generally regarded as a problem.

--
  Phlip
    http://www.c2.com/cgi/wiki?TestFirstUserInterfaces