
When I was working at Thiokol I found an article online about the
development process for the software that runs on the shuttle. As you can
imagine they have a very careful development process. I'll see if I can
find it.
I found your blog entry by following a link from my referrer log. I think
there's plenty of open-source tools for testing, but testing itself is seen
as a second-hand activity by most non-agile shops. Testing does get first
class citizenship in agile (i.e. XP, Scrum, etc.) environments, although
there's a big emphasis on unit tests there and less on
system/integration/load/performance/stress tests.
Grig-
Thanks for the comments. I guess I should have clarified what I was saying
by "end users". My bad. I meant your ISV/fortune 1000 type companies
using open source, who often times will have a "performance engineer" in
house.
A few years ago I decided I wanted to contribute to development of free
Common Lisp implementations (there are a half dozen or so of these.)
Looking at what they needed, I decided testing was the biggest lack.
> developers are more willing to make daring, dangerous changes knowing
that testing is more likely to catch regressions.
The Perl community, and the CPAN in particular, has a very strong culture
of testing. Older CPAN modules sometimes tend to have less tests, but the
newer modules (and some heavily used modules like DBI, etc) tend to have
very good tests suites. There is even a CPAN Testers group which will run
these automated test suites on multiple platforms and report back to the
module authors on their results. I would suspect that a lot of this comes
down to the fact that Perl has an excellent assortment of testing
tools/modules available and the community itself places a high value on
using these tools.