Re: TAP::Harness timeout?

On Wed, Jan 29, 2014 at 6:50 AM, Todd Rinaldo <[email protected]> wrote:

> I’m looking at using TAP::Harness to process our test suite. Up to now
> we’ve been using some home grown code that IMHO is a heroic attempt to
> re-implement TAP::Harness.
>
> It seems to do everything we need with one exception. We have rules that
> disallow a unit test from taking more than XXX seconds to run. If it
> exceeds that, we abort the test and move on to the next one, declaring the
> long test file a failure. Before I look at subclassing, can anyone tell me
> if there’s a way to do this with the existing code?
>

I don't think there's a clean way of doing that. You could hack something
up with alarm, but that wouldn't work well with parallelization (or
multi-threading for that matter). Another possibly route is subclassing
TAP::Parser::Iterator::Process to use a timeout (not that IO::Select makes
it easy to do that correctly, given crappy error signaling), and you'd have
to do the same in TAP::Parser::Multiplexer if you want parallelization.

Leon