From karels Mon Oct 21 14:09:27 1985
Received: by monet.berkeley.edu (5.29/1.2)
	id AA02031; Mon, 21 Oct 85 14:09:13 PDT
Date: Mon, 21 Oct 85 14:09:13 PDT
From: karels (Mike Karels)
Message-Id: <8510212109.AA02031@monet.berkeley.edu>
To: miriam
Subject: guessing games?
Status: R

Is sendmail now engaging us in guessing games?  I just got this back
after sending to a mailing list; I wonder which of the 30 addresses is bad.

		Mike

------- Forwarded Message

Received: by monet.berkeley.edu (5.29/1.2)
	id AA01352; Mon, 21 Oct 85 12:03:32 PDT
Date: Mon, 21 Oct 85 12:03:32 PDT
From: MAILER-DAEMON (Mail Delivery Subsystem)
Subject: Returned mail: Host unknown
Message-Id: <8510211903.AA01352@monet.berkeley.edu>
To: karels

  ----- Transcript of session is unavailable -----

   ----- Unsent message follows -----
Received: by monet.berkeley.edu (5.29/1.2)
	id AA01342; Mon, 21 Oct 85 12:03:32 PDT
Date: Mon, 21 Oct 85 12:03:32 PDT
From: karels (Mike Karels)
Message-Id: <8510211903.AA01342@monet.berkeley.edu>
To: beta43_sites
Subject: beta test status

Thanks for all of the feedback so far.  It appears that things are running
fairly succesfully at the sites that I've heard from.  I've summarized
the more useful bug fixes here, so not everyone will have to deal with the
same problems.

Incidentally, some of the bug reports have been sent using sendbug
to 4bsd-bugs@ucbarpa.  As much as I hate to discourage using sendbug,
I would prefer that beta test bugs be sent to beta43_bugs@monet.berkeley.edu
or ucbvax!beta43_bugs@monet.  Sendbug could be modified to use that address
for now if you prefer.

I would like to get status reports from the sites that haven't yet sent
anything, just to have an idea of how far the testing is proceeding.
Thanks for all of your help.

		Mike


1. There were two versions of sendmail on the tape, both with some problems.
Eric sent a message about this earlier with the fix for the one in /usr/src.
The newer, more experimental version on the end of the tape has been updated,
and is available by anonymous ftp from ucbarpa in pub/sendmail.tar.

2. The binary of /usr/ucb/mail was mysteriously broken.  Recompilation fixes
the problem.

3. There was a bug fix for config in the printed change list; unfortunately
it was incomplete.  The problem was multiple inclusions of files like if_uba.o
which have multiple entries in files.* with different dependencies.
A diff containing the complete fix follows; it includes an unrelated change
to allow .o files to be configured:
8c8
< static char sccsid[] = "@(#)mkmakefile.c	5.2 (Berkeley) 9/17/85";
- ---
> static char sccsid[] = "@(#)mkmakefile.c	5.5 (Berkeley) 10/14/85";
34c34
<  * Lookup a file, by make.
- ---
>  * Lookup a file, by name.
192c192
< 	register struct file_list *tp;
- ---
> 	register struct file_list *tp, *pf;
238c238
< 	if ((tp = fl_lookup(this)) && (tp->f_type != INVISIBLE || tp->f_flags))
- ---
> 	if ((pf = fl_lookup(this)) && (pf->f_type != INVISIBLE || pf->f_flags))
288c288,289
< 	tp = new_fent();
- ---
> 	if (tp == 0)
> 		tp = new_fent();
323c324,325
< 	tp = new_fent();
- ---
> 	if (tp == 0)
> 		tp = new_fent();
334a337,338
> 	if (pf && pf->f_type == INVISIBLE)
> 		pf->f_flags = 1;		/* mark as duplicate */
448a453,456
> 	if (och == 'o') {
> 		fprintf(f, "%so:\n\t-cp ../%so .\n", tail(np), np);
> 		continue;
> 	}

4. The change to syslog priorities didn't get made in the bbnnet directory.
Each occurrence of KERN_RECOV should be changed to LOG_INFO.

5. The binary of arp was out of date.  Recompilation will fix it.
This was due to a change in <netinet/if_ether.h>.

6. There was a bug in ftp, finger and probably other programs that tried
to fill in a hostent structure when given an address in dot notation
rather than a hostname.  The diff for ftp was sent earlier, but is included
here; the others can be fixed similarly:

Oct  5 18:40 1985  SCCS/s.ftp.c: -r5.1 vs. -r5.2 Page 1

49a50
> 		static char *addrbuf;
59a61
> 		def.h_addr_list = &addrbuf;

7. The driver for the Proteon ring interfaces (vaxif/if_vv.c) was omitted
in error.  If anyone who needs it hasn't gotten a copy, let me know and I'll
send it.

8. There was an obscure problem with TCP (netinet).  Under certain
circumstances, it would incorrectly consider the window closed and refuse
to accept data.  Under these conditions, FTP, etc. would create a zero-length
file and then wait forever.  The fixes are as follows:

Oct 14 17:43 1985  SCCS/s.tcp_seq.h: -r6.2 vs. -r6.3 Page 1

25c25
< 	(tp)->rcv_nxt = (tp)->irs + 1
- ---
> 	(tp)->rcv_adv = (tp)->rcv_nxt = (tp)->irs + 1


Oct 14 17:41 1985  SCCS/s.tcp_input.c: -r6.13 vs. -r6.14 Page 1

329c329
< 	tp->rcv_wnd = MAX(sbspace(&so->so_rcv), tp->rcv_adv - tp->rcv_nxt);
- ---
> 	tp->rcv_wnd = sbspace(&so->so_rcv);
331a332
> 	tp->rcv_wnd = MAX(tp->rcv_wnd, (short)(tp->rcv_adv - tp->rcv_nxt));



------- End of Forwarded Message