Running t/best.t takes too long about algorithm-networksort HOT 2 CLOSED

At the moment, you shouldn't be running the full extent of the tests at all.

In the currently released version, the tests only go to a small value, for the sake of the testers. It's the developers who can get the long (hour+) test (believe me, I know, I had to catch a couple of errors with the new networks in #8).

I wrote my own method of handling that, but there's a standard way of separating "install" tests from "developer" tests that's either new or which I missed the first time. I'll be upgrading the Build.PL file to make use of it.

In the mean time, just don't use "build test". If you don't feel comfortable with that, use

build test test_files=t/00-load.t

to at least check for compilation errors.

from algorithm-networksort.

Hehe roger that on the hour+ test times...

I was just looking at master on github, good to hear it will be pared down for the CPAN release, thanks.

PS. I've considered making an Algorithm::Networksort:ZeroOne::XS module before that does this testing in parallel with optimised C code. Someday... :)

from algorithm-networksort.