Path: utzoo!utgpu!jarvis.csri.toronto.edu!rutgers!mailrus!iuvax!bsu-cs!dhesi
From: dhesi@bsu-cs.bsu.edu (Rahul Dhesi)
Newsgroups: comp.unix.questions
Subject: Re: RTFM... (really: sed -n 100,115p)
Message-ID: <7330@bsu-cs.bsu.edu>
Date: 19 May 89 22:58:54 GMT
References: <2859@buengc.BU.EDU> <611213677.29680@ontmoh.UUCP> <1031@philmds.UUCP> <7298@bsu-cs.bsu.edu> <1033@philmds.UUCP> <7320@bsu-cs.bsu.edu>
Reply-To: dhesi@bsu-cs.bsu.edu (Rahul Dhesi)
Organization: CS Dept, Ball St U, Muncie, Indiana
Lines: 25

In article <7320@bsu-cs.bsu.edu> I wrote:
>>Are you
>>implying that each filter should read all its input ??
>
>Yes, unless it knows that a process providing that input is prepared to
>gracefully handle the situation when writing to stdout causes an
>error.

Reconsidering this, I realize that the SIGPIPE that occurs when the
process writes to a broken pipe ought to abort it silently.  The user
will not see the "broken pipe" message unless the process catches
SIGPIPE and prints the message.

Unfortunately, I have come across programs that do catch SIGPIPE and
print an error message.

Further, it's quite conceivable that a process that is writing to its
standard output will (a) finish writing and then (b) do some cleanup.
Unless such a process catches SIGPIPE, it will simply abort without
cleaning up.

So I still think that filters should read all their input.
-- 
Rahul Dhesi <dhesi@bsu-cs.bsu.edu>
UUCP:    ...!{iuvax,pur-ee}!bsu-cs!dhesi