Boost logo

Boost :

From: Beman Dawes (bdawes_at_[hidden])
Date: 2002-09-04 10:15:18


At 05:02 AM 9/4/2002, Gennadiy Rozental wrote:

>Responding to requests for lightweight testing I implemented header
>boost/test/minimal.hpp.

Excellent! Thanks! I'm running the regular regression tests now; as soon
as they finish I'll try boost/test/minimal.hpp on one or two.

Nits:

* The correct spelling is "deprecated"

* boost/test/minimal.hpp is missing a copyright message.

> Using of this header is similar to the using of Boost.Test v1. you
>including the header and implement test_main function. Minimal test keeps
>track on number of error and provide unified error logging in case of
both
>fatal and non fatal errors. BOOST_INCLUDE_MAIN is not used. main is
>included unconditionally.

For backward compatibility, please considered adding something like the
following to <boost/test/test_tools.hpp> after the include guards:

# ifdef BOOST_INCLUDE_MAIN // workaround to keep old code from breaking
# include <boost/test/minimal.hpp>
# else
     ... current contents
# endif

That makes it easier for regression testers who are running both the old
and new versions of the Boost regression tests. This includes some
compiler suppliers who use the Boost tests as part of their own tests.

You wouldn't have to document this; it is pure workaround.

> One consequence is that minimal test could not be
>used for multi-unit testing. You will need to use other Boost.Test
>components for that.

That seems perfectly reasonable to me.

>Also if you did not remark it from one of the messages in some unrelated
>queue, there is another addition. To support library-less usage of
>Bosot.Test v2 components I introduced
>3 new headers in subdirectory boost/test/included:
>boost/test/included/prg_exec_monitor.hpp
>boost/test/included/test_exec_monitor.hpp
>boost/test/included/unit_test_framework.hpp
>
>If you use these headers you are agree with possibly some compile time
>overhead but you won't need to link with precompiled component's
library.

Good!

>Comments?

Nice progress. Looks like you are about done.

>P.S. Should I try to reflect this changes in docs before the release?

Yes. I told someone working on some other docs to try for a Friday check
in, but I don't know if they will make that schedule.

--Beman


Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk