On Wed, Jul 23, 2008 at 10:33:15PM -0700, Paul Makepeace wrote:
> On Wed, Jul 23, 2008 at 8:26 PM, Lyle <[EMAIL PROTECTED]> wrote:
> > That's interesting. From Perl Cert posts I've read on LPM and Perl Monks
> > a lot of people don't rate Brainbench tests. The main argument being
> > that the tests are based on memory of functions and their parameters, as
> > opposed to whether you can actually write code.
> I haven't read the posts so was wondering if they had an opinion on
> how you would test whether someone could write code? I can see that's
> kinda testable in an interview but doing it in a certifiable way seems
> tough to me.

Here at $work we test people by having them come in for a programming
test.  We give them an hour and a half with perl and an interwebnet
connection to solve as much as they can of a fairly open-ended problem.
The problem tests knowledge of algorithms and data structures, and the
really good people get enough done that they can start on optimising
their code.

In the interview we go through the code together, discussing other
approaches they might have taken, why they chose particularly methods,
and also talk about testing and documentation (which we don't expect
them to do in the test itself).

Yes, it's hard to certify - although don't universities and schools
manage to do it in computer science courses? yes they do, but to do so
is expensive - but we don't *need* that.  We need people who can
convince me and my colleagues that they're competent.

-- 
David Cantrell | Godless Liberal Elitist

You can't spell "slaughter" without "laughter"
_______________________________________________
BristolBathPM mailing list
[email protected]
http://mailman.bristolbath.org/mailman/listinfo/bristolbathpm

Reply via email to