apr-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Ryan Bloom" <...@covalent.net>
Subject New test suite, decision made.
Date Fri, 12 Jul 2002 17:02:41 GMT

I have made my decision for the test framework that I will use for the
new APR test suite.  It will be cutest.  I will include a copy of cutest
in the CVS repository, so that people who don't want to download a
special program won't have to.  Cutest is zlib licensed, so it won't
have any conflict with our license.

At the end of the day, it came down to two test frameworks, Subversions
and CuTest.  Both have advantages and disadvantages, but Subversion's
test framework really doesn't do enough as far as creating a
comprehensive system that we can use.  Their setup is a main driver
program that can call individual test libraries.  It is missing some key
features, such as consolidating tests, verbose output, a summary output,
and looping tests.  CuTest is also missing all of these features

Since adding all those features to either project would be relatively
easy to do, it is a judgment call about where it should be done.  I have
decided to add these features to CuTest.  This is because CuTest is a
project that is specifically designed to create a C test suite.
Subversion isn't a test suite project, the fact that it happens to have
a test suite is a bonus.

Rather than add features to a test suite that isn't meant to be a full
test suite, I would rather add features to a project that is trying to
be a full test suite.  After discussing this with Ben Collins-Sussman on
the svn IRC channel this morning, I believe that he agrees.  If we are
going to add features to the test framework as we discover that we need
them, then we should use a real test framework.

Once that decision is made, we have the option to create our own test
framework.  I am not willing to re-invent the wheel on this.  We tried
to do that once, but we aren't test framework developers, and the
hodge-podge that is APR's current test suite proves that.

So, that is my reasoning for using CuTest as our test framework.  I am
more than happy to be told that I am wrong and that we should use some
other test framework. 


Ryan Bloom                  rbb@covalent.net
645 Howard St.              rbb@apache.org
San Francisco, CA 

View raw message