On 5 Oct 2006, at 15:11, Paul Beckingham wrote:
Recently I was required to create another flavor of test harness
that runs tests, then captures and stores output.
The nature of my testing means that I am running millions of tests,
and the resultant captured output is therefore huge. So I modified
my tests to do the *equivalent* of the following:
[snip]
Personally I would say it would be better to leave the things
generating the TAP output alone, and have the test harness only store
the test failures. Ovid's TAPx::Parser makes this pretty darn easy,
and it's not stupidly difficult with Test::Harness::Straps although
it needs some sneakyness with private methods.
Adrian
