Re: [bug-gawk] Behavior of fflush with SIGPIPE on stdout [PATCH]

[Andrew J. Schorr]

On Thu, Mar 23, 2017 at 01:36:40PM -0600, address@hidden wrote:
> I think this patch addresses all your issues.

Perhaps, although flush_io calls is_non_fatal_redirect before there's
an I/O error. I still dislike doing that lookup before anything bad happens.

It's also not obvious to me whether calling update_ERRNO_int(errno) after
flush_io returns is going to work. Is it possible that flush_io will make
other calls after the failed call that could reset errno?

Regards,
Andy