Paul Eggert <[EMAIL PROTECTED]> wrote: ... > My kneejerk reaction is that a modified 'closein' is a better way to > go here. That is, the affected programs (and there are not that many > of them) should invoke close_stdin if they have attempted to read > anything from stdin and have not read end-of-file.
I've been rethinking my position here, too. Another reason to prefer something like closein is the gnulib philosophy of penalizing only losing systems. It's a shame that modern glibc is in the "losing" camp, on this.
