platforms problems

[Stefan Seifert]

On Friday 03 August 2012 16:08:15 Moritz Lenz wrote:

> There's a very fundamental question underneath: If you declare that you
> support a platform, you don't just TODO failing tests for it when
> introducing a mostly unrelated feature. If you do, the support for the
> platform is second class, ie you don't care about that platform as much
> as about Linux.

A mostly unrelated feature that can be turned off. When compiled --without-
threads, Parrot falls back to using usleep instead of the timer thread. So 
it's not even a matter of losing a feature on a platform but of having to 
chose between two features. It might very well be possible that previously 
existing features continue to work with just threading not being available.

Of course someone would have to test this. I'd still very much like to see 
threading work on all supported platforms. But I'll need access to them to find 
out what's wrong. Or someone with access would have to dig into the timer stuff 
to find out. It's actually not very complicated.

Stefan