Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.08 10/3/83; site psuvax1.UUCP
Path: utzoo!linus!decvax!mcnc!akgua!psuvax1!jdi
From: jdi@psuvax1.UUCP (John D. Irwin)
Newsgroups: net.bugs.2bsd
Subject: Re: 2.9 uucp : beware!
Message-ID: <1117@psuvax1.UUCP>
Date: Thu, 2-Aug-84 12:46:55 EDT
Article-I.D.: psuvax1.1117
Posted: Thu Aug  2 12:46:55 1984
Date-Received: Sun, 5-Aug-84 07:57:24 EDT
References: <2097@usceast.UUCP>
Organization: Pennsylvania State Univ.
Lines: 35

He He.  This reminds me of my experience.  I brought up 2.9 about two weeks ago
(which was really fun considering we have no tape drive on our 11/34) and tried
to transfer a 7M hunk of source over from the vax to 2.9.  Well, I didn't come
in the next day but found out later that in 23 hours it had done something like
2.9M -- ie: about 30 cps over a 9600 baud line.

Anyway, after cursing some I looked at the packet driver code and found this
little interesting piece in the packet get routine:

	alarm(PKTIME);
	for (nchars = 0; nchars < n; nchars += ret) {
		ret = read(fn, b, n - nchars);
		if (ret == 0) {
			alarm(0);
			return(-1);
		}
		PKASSERT(ret > 0, "PKCGET READ", "", ret);
		b += ret;
		sleep(1);
	}
	alarm(0);
	return(0);
}


Isn't that cute?  A one second sleep for every packet!

Well, I took out the sleep, but performance was still only 300 cps or so so I 
just installed 4.2uucp which has worked fine since.

-- 
Spoken:	John D. Irwin
AT&T:	814-237-5068
Nets:	jdi@psuvax1.{BITNET,CSNET}
Uucp:	{akgua, allegra, cornell, pitt, purdue, ihnp4, burdvax}!psuvax1!jdi