On Thu, Feb 10, 2005 at 01:56:50PM -0800, chromatic wrote: > On Thu, 2005-02-10 at 16:46 -0500, Michael G Schwern wrote: > > > Trouble is STDOUT is parsed while STDERR is not so T::H has to be able > > to disambiguate them while still displaying them in the correct ordering > > and not run afoul of any buffering issues. > > I wave the "Nope, the UNbiquitous Test::Harness::Straps" flag in hopes > that it staves off the backwards compatibility police.
Err, I don't get it. > > And no, users should not have to upgrade their OS nor add sequence numbers > > to output. > > Yeah, but they *should* upgrade to operating systems that implement > *some* form of usable process control. I'm playing my idealist card > here and checking that we're still only betting peanuts and > marshmallows. I reference RFC 1925, section 2.1 "It Has To Work". If its a platform p5p considers supported, Test::Harness will work. Perhaps not all the fancy features, but it will parse and display test results. So as long as TH's basic functionality continues to work everywhere, that's fine.