> There ought to be a way to capture diagnostic information in the TAP stream > because it's useful for diagnosing problems.
I feel like I'm talking to myself when I say this (since I've said this before) but I'll say it again just, well, because :) the implicit idea that STDERR generally goes to the screen is simply not always the case - if you produce your TAP in an environment outside of the shell (for example, in an httpd server, mail server, across ssh, whatever) then STDERR is likely the remote environment (error_log, syslog, etc). I'm _constantly_ bitten by diag() foo going to someplace other than my test output in these cases. the more popular the TAP stuff people are building becomes, I suspect you'll see a greater percentage where the whole STDOUT/STDERR isn't at all what is important - it's TAP output versus platform runtime errors (like perl warnings in scripts). so, to chromatic's point, I can't help but feel like solving the quoted issue would go a long way toward reducing the costs in this debate. --Geoff
