🚀 go-pugleaf

I've been using ppp *over* rlogin recently, and I
encountered a rather interesting situation when the modems
hung up by themselves at one point.

I'm using the ubiquitous pl6 ppp, although from looking at the
code I think that bkc's 1.1 would behave similarly.

Here's how I set up the ppp link: home system (home) calls work system
using a T3000. Work system 1 (work1) answers. Log in as some user, then
run rlogin -E -8 to a system that has ppp installed (work2), and specify
a password entry that starts up ppp in server mode.

If, for some reason, the modem connection dies, SIGHUP gets sent
to the client rlogin running on (work1). This causes the rlogin client
to exit.

Presumably (I'm speculating from this point on), this will cause
SIGPIPE to be sent to the server ppp running on (work2) at some
point. From the code, it seems that this signal is currently ignored.
In any case, the ppp server on (work2) continues running after the
client rlogin has gone away.

I propose that the server ppp ought to catch SIGPIPE and interpret
it the same way as it currently interprets SIGHUP. Is this a reasonable
way to handle the above scenario?

Is there any reason ppp should NOT handle SIGPIPE like this always?

go ahead, enlighten me :-)

 ~!paul
-- 
Paul Ziemba  paul@alantec.com	alantec!paul  408.955.9000 x227

"The artist is not a special kind of man, but every man is a
special kind of artist." -- Ananda K. Coomaraswamy