On 29 Dec 2007, at 23:31, Ovid wrote:
Hi all,
I've just released a new version of Test::Aggregate
(http://search.cpan.org/dist/Test-Aggregate/).
[snip]
A quick experience report.
After about three hours work I shifted all of the tests that worked
without changes in a $work project to use T::A.
* Core tests run time drops by about 30% (wall clock).
* Found two new bugs in the code under test. One due to some bad
stateful interaction that showed up once a bunch of tests were
running under T::A. One deadlock issue that showed up once we were
running the stress tests more quickly.
* Pointless warnings from UNIVERSAL::can needed to be stomped.
* The Perl::Critic tests slowed the T::A tests down. Didn't bother to
figure it out - just swapped it out again. Anybody know why?
* The tests that didn't go across (apart from the P::C stuff) are
most because of bad design decisions (in the code or the tests). Most
of which I'd spotted before :-) It proved a nice way of highlighting
them.
Considering the number of times I run those tests at the moment that
three hour investment will be paid off very quickly. So.... result!
(as far as I'm concerned :-)
Ta Ovid :-)
Cheers,
Adrian
