Ovid wrote: > You've studiously avoided answering whether or not you would accept a patch > for Test::Builder > which would allow STDERR to be sent to STDOUT. I realize Test::Harness can't > use it, but others > can. Unless I'm missing something very fundamental, it makes most of my > problems in this area > just Go Away. I can try to write a bunch of complicated code -- as you > suggested with "some > combination of IPC::Run, IPC::Run3 and IPC::Open3" -- but solving the problem > at the source is > far more correct than trying to fix it after the fact.
What's the old Unix adage: Be strict in what you emit, lax in what you accept. Nobody said writing a test harness was going to be easy. > If there is some technical reason why this approach won't work, please let me > know. I don't > see the problem when reviewing the Test::Builder code. I know I've replied on this at least once back when the whole "let's cross the streams" discussion first came up and said it won't work. http://www.mail-archive.com/perl-qa@perl.org/msg06694.html I'd written a second one a couple days ago but forgot to send it, I just did. My parents are in town visiting so I don't have a lot of time at the moment. Let me sum up how I see it: 1) DON'T PANIC! Test::Harness has lived for nearly 15 without dealing with STDERR. 2) I haven't heard why this is an urgent MUST-HAVE-HACKY-FIX-NOW-NOW-NOW issue. Seems to me you've hit your first cross-platform compat bump. There will be more. 3) It breaks TODO tests. 4) Test::Builder is not the only game in town. 5) Even if it was, people ship with specific (ie. old) versions of Test::Builder in t/lib. Its not urgent. It won't work. It doesn't solve your problem. Is there something I'm missing here? Is TAPx::Parser being used in production? I'm being resistant and conservative in my normal way when it comes to toolchain stuff. I'll have more time for this on the weekend.
