Jeff Trawick wrote: > > Users occasionally run scripts that spew noise to stderr that may not > require logging except when debugging. Perhaps a more general > FcgidChildStderr directive could enable both the legacy Win32 > processing or other solutions.
I'd rather 1. set the expectation that stderr channel exists (even if it's /dev/null'ed - oops - I mean \Device\NUL'ed ;-) as the defacto standard, just as on unix and 2. focus on how to tune the log level, following the pattern of mod_cgi[d] and adjusting those again, accordingly. Remember that stderr is rewired through the Fcgid processing phase. We are only talking here about stderr-at-startup.