RE: [boost] Draft of Test Policy web page

From: Beman Dawes (bdawes_at_[hidden])
Date: 2001-01-30 10:45:34

At 09:37 AM 1/30/2001 +0000, Moore, Paul wrote:

 >From: Beman Dawes [mailto:bdawes_at_[hidden]]
 >> There is a draft of a Boost Test Policy and Protocols page up at
 >> http://www.boost.org/more/test_policy.htm.
 >>
 >> (This page isn't "live" yet; there are no links to it from
 >> elsewhere on the
 >> site, and the links on the page won't all work.)
 >>
 >> It includes a bit of speculation - references to Boost test related
 >> libraries that haven't been accepted yet.
 >>
 >> Comments appreciated.
 >
 >What about compile-time failures? Suppose I want to test that a certain
 >construct *won't* compile? (The obvious case would be some form of
 >compile-time assert). There seems to be nothing covering that...

Hum...  regression.cpp can handle at least six commands.

   * compile only, expecting success.
   * compile only, expecting failure.
   * compile and link, expecting success.
   * compile and link, expecting failure.
   * compiler, link, run, expecting success.
   * compiler, link, run, expecting failure.

I wouldn't be surprised if one or two more get added.

The issue is how and where to document regression.cpp.  It isn't a formally 
reviewed boost library; it is really just an internal tool, so we don't 
want to oversell it.  OTOH, it is really useful, so we don't what to hide 
it either.

That's getting away from you point, however.  I'll add something to the 
Test Policy page indicating that testing isn't limited to just "run" tests.

Thanks,

--Beman