Eric Blake-1 <[EMAIL PROTECTED]> writes: > I think you meant <errno.h>, not <error.h>. The latter is glibc > specific, and needs explicit testing (for example, cygwin does > not provide an error.h).
I meant <error.h>, actually. Does Cygwin have an error_at_line function without having an <error.h>? If not, then it shouldn't be an issue, as the test is for error_at_line.
