Sun-Spots-Request@RICE.EDU.UUCP (06/04/87)
SUN-SPOTS DIGEST Thursday, 4 June 1987 Volume 5 : Issue 17 Today's Topics: bug in make(1) OOPS IGNORE the "WARNING: about Sun SCSI shorting" Re: mouse problems Re: 'textedit' tool? 4.3 acucntrl for SUN Re: query on dump to sun scsi tape? Re: taking system down for dumps other than level 0? Re: Request for sendmail.cf (very long) Make your workstation a bit more secure Sun workstations for the home? 3rd party disk drives? Color Hard Copy Units? Clean version of dbxtool? sun's ie connected to a DECOM? 3.0/3.2 .rhosts and 3.2 diag? Function key programming? ------------------------------------------------------------------------ Date: Thu, 28 May 87 20:54:23 PDT From: fritz@vlsi.caltech.edu (Fritz Nordby) Subject: bug in make(1) Under some conditions, make(1) sets the environment variable IFS to an empty string. This almost guarantees that Bourne shell scripts invoked by make will die in bizarre ways. To repeat, create an empty directory, cd there, and do: % /bin/sh $ cat >cc <<FOO #!/bin/sh echo \$# set "\$@" echo \$# FOO $ chmod +x cc $ cat >makefile <<FOO foo: bar cc foo bar spam FOO $ >bar $ PATH=:$PATH $ IFS=' ' $ # that's IFS='<space><tab><newline>' $ export PATH IFS $ cc foo bar spam 3 3 $ make cc foo bar spam 3 1 $ ^D % The error is the final 1; the set "$@" in cc is expanded to set "foo" "bar" "spam" which, after quote processing, becomes set foo bar spam However, since IFS is an empty string, there are NO FIELD SEPERATORS, so this sets one positional argument; namely, this is equivalent to set 'foo bar spam' This results in the anomalous "1" output by cc when invoked by make. The only quick fix I can think of is to have all my shell scripts set IFS explicitly, and that's a massive kludge, not a fix. Fritz Nordby. fritz@vlsi.caltech.edu cit-vax!fritz ------------------------------ Date: Tue, 26 May 87 22:10:30 edt From: cmcl2!seismo!decvax!savax!dove@hp-pcd.UUCP (Webster Dove) Subject: OOPS IGNORE the "WARNING: about Sun SCSI shorting" It seems some guys here took the SCSI apart and forgot to reinsert the plastic shields that prevent the RF fingers from shorting the ribbon cable at the outer connector. My apologies for the mistaken information. Yr Hmbl Svt Web Dove ------------------------------ Date: Fri, 29 May 87 00:20:02 PDT From: kevin@sun.com (Kevin Sheehan) Subject: Re: mouse problems The problem is that the driver get into a state where it deletes null characters from the stream. Pulling the mouse causes the driver to reset. Is fixed. ------------------------------ Date: Fri, 29 May 87 09:36:25 PDT From: ultra!wayne@ames (Wayne Hathaway) Subject: Re: 'textedit' tool? I too tried to get 'textedit' to insert a special character into a file (in my case a ^G, or BEL). After fighting the documentation and experimenting for a while, I called the Sun Hotline. While they were very friendly and responsive, the bottom line answer was "no way." (I did find a way, by using 'cat' to create a file with just a BEL in it and having 'textedit' "get from file" it into the appropriate place.) My real solution? Use the ^Q "quoted-insert" mechanism -- in GNU Emacs! Wayne Hathaway ultra!wayne@ames.arpa Ultra Corporation 2140 Bering Drive San Jose, CA 95131 408-922-0100 ------------------------------ Date: Fri, 29 May 87 18:33:01 EDT From: budd@bu-cs.bu.edu (Philip Budne) Subject: 4.3 acucntrl for SUN As has been said, you shouldn't need acucntrl with the Sun ttyd0/ttym0 dualism. However, I did it anyway, having had some trouble playing with the devices. Is acucntrl public source? Phil Budne Boston University ------------------------------ Date: Sat, 30 May 87 15:52:56 From: H_Eidnes%vax.runit.unit.uninett@tor.nta.no (Havard Eidnes) Subject: Re: query on dump to sun scsi tape? In Sun-Spots v5n16 Michal Khaw writes: > What flags and parameters (other than 'c') are recommended for doing dumps > to sun3 scsi cartridge tapes? Should I try to specify tape length, density, > blocking, and if so, what values should I use? Is /dev/rst0 or /dev/rst8 > preferable? On our Suns, we have the tape drives that are both capable of handling QUIC-11 and QUIC-24, the latter being higher density. If your tape drive handles both of these, they are respectively /dev/rst0 and /dev/rst8. We found that using only the "c" argument to dump didn't work very well. Maybe it worked with the QUIC-11, 300ft tapes, but we use 450ft tapes and QUIC-24. I use the following command to dump our file systems: dump ${level}cunbsdf 126 450 10000 /dev/rst8 ${filesystem} This almost fills up the entire tape. If you're afraid of writing too far, you may reduce the density (10000) number. Please note: The above information is something I found by experimenting. I found the SUN documentation somewhat lacking in precision on this point. I think I first got the 126 number from an installation instruction for Sun-Unix 2.0. If some of these parameters should be changed, I'm just happy to get to know the correct ones. ------- E-Mail: <h_eidnes%vax.runit.unit.uninett@nta-vax.arpa> (or @tor.nta.no) H}vard Eidnes (or TeXish: H\aa vard Eidnes) Division of Computer Science, Norwegian Institute of Technology ------------------------------ Date: Sun, 31 May 87 03:00:38 EDT From: dpz@aramis.rutgers.edu (David P. Zimmerman) Subject: Re: taking system down for dumps other than level 0? David, We have 15+ Unix boxes that we do level 0s weekly and level 1s daily on, while the machines are up multiuser. A few are by necessity, one being in a galaxy far far away on the other end of a T1 line, but most are just because it is convenient to not have to take all of those machines down. We have no problems with backups on those live systems and restores from those backups. I wouldn't think that anything besides an active file would be at risk, if even that. dpz ------------------------------ Date: 1 Jun 87 22:29:23 GMT From: richl%penguin.uss.tek.com@relay.cs.net (Rick Lindsley) Subject: Re: Request for sendmail.cf (very long) > From: arnold%umn-cs-fsa@umn-cs.arpa (Doug Arnold) > Subject: Request for sendmail.cf which works with subdomain style hostnames? > > Can anyone provide a sendmail.cf file which works on a Sun host > running 3.2 with a four part (subdomain style) hostname (like > s1.ima.umn.edu)? I want to be able to send and receive mail from > machines on the local ethernet directly and on the internet via a (non > Sun) mailhost which is on the local ethernet. My request to Sun > elicited the following response: the supplied versions of > sendmail.cf don't work for this type of hostname and require quite > extensive modifications; such modifications have been successfully > carried out, but they don't know by whom; why don't I send an appeal over > Usenet. Here is a config file which works. It is used on my workstation (penguin.uss.tek.com). It too will need to be modified, but you should find it easier to modify to your purposes, I think. It is fairly stupid, in that it will send directly to all sites in my tertiary domain, but forward everything else to a "smart" gateway. (See the classes section.) You can make it smarter by enlarging Ruleset 0. Sendmail experts may quibble over the choices of representation used here and may also find some inexplicable chaff in places. This is probably because this is just a smaller version of the version running on tektronix, which is much more complex and much larger. We here at Tek have between 15 and 20 subdomains (or 'tertiary domains', as I've renamed them at points below), and this config file or something similar to it works everywhere, whether a site is a gateway or simply a leaf. The rewriting rules have been rewritten (!) because while they may have been all that Berkeley wanted, for us they were alternately insufficient and overkill. Even non-complex sites may find some of the newly rewritten rules useful. Rick Lindsley richl@penguin.uss.tek.com ...!tektronix!penguin.uss.tek.com!richl ################################################### # Macros # # L top level domain name # # D second level domain name # # I tertiary domain I am in # # F gateway to forward to # # U The name I wish to tack on my uucp mail # # V version number # # e how smtp should greet people # # j my official hostname (host.domain) # # n the name of the daemon (for errors) # # l the format of the Unix style From_ line # # o the set of "operators" (component # # seperators) in addresses # # q default format of sender address # # # ################################################### # the following steps around a sun sendmail bug Dwpenguin DLCOM DDTEK DIUSS DFtektronix DU$w De$j Sendmail $v/$V ready at $b Dj$?I$w.$I.$D.$L$|$w.$D.$L$. DlFrom $g $d DnMAILER-DAEMON Do.:%@!^=/[] Dq$g$?x ($x)$. DV6.19 ################################################### # Classes # # w known names for our local host # # L top level domains which I am already # # a part of # # R known smtp hosts within our subdomain # # S known TEK domain hosts (for direct # # sending) # # U names of our local uucp neighbors # # # ################################################### # the last two provided for compatibility CLCOM UUCP CSNET FR/usr/tek/lib/mail/sendmail/uss.hosts FS/usr/tek/lib/mail/sendmail/tek.hosts # we have none #FU/usr/lib/uucp/L.sys %[^ #] Cwpenguin ################################################### # Options # # A alias file # # a time to wait for alias file to finish # # rebuilding # # B substitute this character for blanks # # c it's ok to connect to expensive # # mailers # # d indicate the default delivery mode # # (one of these) # # i interactively # # b background # # q just queue the message # # D rebuild database if it is out of date # # F temp file mode # # g default gid for mailers to run as # # H location of help file # # L default log level # # N default network name # # o assume that space may be delimiting # # names # # Q location of spool directory # # r time to wait before timing out on # # mail transfers (violates protocols) # # S location of status file # # s be sure everything is queued before # # starting transmission # # T return mail after this amount of time # # t time zone names (only referred to when # # running V6) # # u default uid for mailers to run under # # W Wizard's password (encrypted) # # x load average above which we queue # # messages, rather than immediately # # delivering them # # X load average above which we refuse # # connections # # # ################################################### OA/usr/lib/aliases Oa10 OB. Odb OD OF0600 Og1 OH/usr/lib/sendmail.hf OL9 ON$D Oo OQ/usr/spool/mqueue Or2h OS/usr/lib/sendmail.st Os OT3d OtPST,PDT Ou1 OW* Ox3 OX5 ##### Ruleset 0 -- rewrite addresses S0 # first make canonical R$*<$*>$* $1$2$3 defocus R$+ $:$>3$1 make canonical R$-<@>$* $#local$:$1 handle special local cases R@ $#local$:$n handle <> form # handle undomained hosts that we know about R$*<@$=R> $#smtp$@$2$:$1<@$2> user@subhost R$*<@$=S> $#smtp$@$2$:$1<@$2> user@tekhost R$+<@$=U> $#uucp$@$2$:$1@$2.UUCP user@uucphost # # at this point, we have stuff<@name> or stuff<@name.domain>, with # possible additional .domain's tacked on after the >. Some of the # domains may be our own ... we must check for validity. A plain <@name> # is something we don't know about, and we'll send it to the gateway. # # first check our local subdomain R$*<@$=R>.$I.$D.$=L $#smtp$@$2$:$1<@$2>.$I.$D.$3 user@subland R$*<@$=R>.$I.$D $#smtp$@$2$:$1<@$2>.$I.$D user@subland R$*<@$=R>.$I $#smtp$@$2$:$1<@$2>.$I user@subland R$*<@$+.$I>.$D.$=L $#error$:Host $2 not listed in $I subdomain R$*<@$+.$I>.$D $#error$:Host $2 not listed in $I subdomain R$*<@$+.$I> $#error$:Host $2 not listed in $I subdomain # resolve names that belong in the other subdomains (one provided as an # example) # ICO R$*<@$+.ICO>.$D.$=L $#smtp$@vice$:$1<@$2.ICO>.$D.$3 user@icoland R$*<@$+.ICO>.$D $#smtp$@vice$:$1<@$2.ICO>.$D user@icoland R$*<@$+.ICO> $#smtp$@vice$:$1<@$2.ICO> user@icoland # resolve names that belong in the TEK domain R$*<@$=S>.$D.$=L $#smtp$@$2$:$1<@$2>.$D.$3 user@tekhost R$*<@$=S>.$D $#smtp$@$2$:$1<@$2>.$D user@tekhost # resolve local UUCP links R$+<@$=U.$+>.UUCP $#uucp$@$2$:$1@$2.$3.UUCP user@host.domain.UUCP R$+<@$=U>.UUCP $#uucp$@$2$:$1@$2.UUCP user@host.UUCP # other names, both domained and undomained, will be kicked upstairs R$*<@$*>$* $#smtp$@$F$:$1<@$2>$3 user@somewhere # remaining names must be local (no brackets) R$+ $#local$:$1 everything else ################################################### # Delivery Precedences # ################################################### Pfirst-class=0 Pspecial-delivery=100 Pbulk=-60 Pjunk=-100 ################################################### # Trusted Users # ################################################### Troot Tdaemon Tuucp Tmmdf ################################################### # Format of Headers # ################################################### H?P?Return-Path: <$g> HReceived: $?sfrom $s $.by $j ($v/$V) id $i; $b H?D?Resent-Date: $a H?D?Date: $a H?F?Resent-From: $q H?F?From: $q H?x?Full-Name: $x HSubject: H?M?Resent-Message-Id: <$t.$i@$j> H?M?Message-Id: <$t.$i@$j> ################################################### ################################################### ## Rewriting Rules ## ################################################### ################################################### ##### Ruleset 1 -- Sender Field Rewriting S1 ##### Ruleset 2 -- Recipient Field Rewriting S2 ##### Ruleset 3 -- Name Canonicalization # -or- # Making Things Useful for Us # # final form when we leave this rule set will be simply # rest-of-addr<@site> # to indicate where this message is going to next. Local mail is indicated # by a lack of angle brackets in the finished product. # S3 # handle "from:<>" special case R<> $@@ turn into magic token # basic textual canonicalization -- note RFC733 heuristic here # if there are angle brackets, they take precedence over anything R$*<$*<$*<$+>$*>$*>$* $4 3-level <> nesting R$*<$*<$+>$*>$* $3 2-level <> nesting R$*<$+>$* $2 basic RFC821/822 parsing R$+ at $+ $1@$2 "at" -> "@" for RFC 822 # make sure <@a,@b,@c:user@d> syntax is easy to parse -- undone later R@$+,$+ @$1:$2 change all "," to ":" # take care of VMS-looking addresses -- we'll make them look uucp-ish # because that's easiest. At the gateway into vms land they will be # properly converted anyway. R$+::$+ $1!$2 # localize and dispose of route-based addresses R@$+:$+ $@$>5<@$1>:$2 handle <route-addr> # more miscellaneous cleanup R$+:$*;@$+ $@$1:$2;@$3 list syntax R$+:$*; $@$1:$2; list syntax R$+@$+ $:$1<@$2> focus on host.domain R$+<$+@$+> $1$2<@$3> look at rightmost host.domain R$+<@$+> $@$>5$1<@$2> already done # convert old-style addresses to a domain-based address R$-:$+ $@$>5$2<@$1> host:user R$+^$+ $1!$2 convert ^ to ! R$-.$+!$+ $@$>5$3<@$1.$2> host.domain!user R$-!$+ $@$>5$2<@$1> host!user R$+%$+ $:$>6$1@$2 user%host R$*@$* $@$>5$1<@$2> NOW we are ready to call 5 ##### Ruleset 4 -- Final Field Rewriting S4 R@ $@ handle <> error addr R$*<@>$=R $?I$1<@>$2.$I$|$1<@>$2$. R$*<@$=R> $?I$1<@$2.$I>$|$1<@$2>$. R$*<$+>$* $1$2$3 remove angle brackets R@$+:@$+:$+ @$1,@$2:$3 <route-addr> canonical # delete duplicate local names R$+%$=w%$=w $1%$3 u%host%host => u%host R$+%$=w@$=w $1@$3 u%host@host => u@host ##### Ruleset 5 -- exclude as many recognizable domains as possible, and # ourselves as well. Only internal tertiary domains will # remain within the <>s. S5 R$*<@$+.$=L>$* $:$1<@$2>.$3$4 exclude own top level domains R$*<@$+.$D>$* $:$1<@$2>.$D$3 exclude our own domain R$*<@$=R.$I>$* $:$1<@$2>.$I$3 exclude subdomain, if member # now exclude the local site, if present # we may easily have a case where the site name and the domain name are # the same -- the next line checks to see if a "." is present # R$*<$*.$=w>$* $@$1<$2.$3>$4 case of domain == site;leave it! R$*<$*$=w>$* $1<$2>$3$4 R<@>:$* $@$>3$1 retry after route strip R$-<@>$* $@$1<@>$2 must be local R$*<@>$* $@$>3$1 strip useless trash & retry ##### Ruleset 6 -- change rightmost % to @. S6 R$*%$* $1@$2 First make them all @'s R$*@$*@$* $1%$2@$3 Undo all but the last. R$*@$* $@$1@$2 Put back the brackets. ##### Ruleset 10 -- Sender address rewriting for local and program mail S10 R@ $n errors to mailer-daemon ##### Ruleset 11 -- convert @s and %s to !s for uucp S11 R$+ $:$>14$1 R$+ $:$U!$1 stick on our host name ##### Ruleset 12 -- convert @s and %s to !s for uucp # some trickery going on here. The $: portion of the mailer (user to mail to) # will also pass through here. We want to strip the destination host from # that name, but we DON'T want to strip it from any other addresses. So # in the $# mailer invocation in ruleset 0, we DON'T put angle brackets # in the name. All # other sites will have had angle brackets added because they passed through # ruleset 3. The last three rules here strip out the destination site, add # angle brackets, re-rewrite the address. In addition, a dummy domain, # .XXXX is added because we don't want ANY @'s in this rewriting, whereas # we've made a policy decision that if a properly formatted user@host # spec comes through, we will leave it alone. # another consideration: how to handle addresses like a!user<@b>. We may # be handed this if the original path looked like "b!a!user". So we have # a special set of rules for that case (ruleset 22). S12 R$+<@$+>$*.UUCP $1<@$2>$3 strip off .UUCP, if present R$*.XXXX $@$>14$1 process user field R$-<@$+>$* $@$1<@$2>$3 if properly domained, don't mess R$+ $@$>14$1 finish processing ##### Ruleset 13 -- handle mixed @'s and !'s (@'s and %'s get precedence) S13 R$+%$+<@$+>$*:::$+ $2<@$3>$4:::$1!$5 R$+<@$+>$*:::$+ $2$3!$1!$4 ##### Ruleset 14 -- convert @s and %s to !s for uucp (!'s get precedence) # it is possible we have the $: user specification from # the mailer invocation (no angle brackets <>) S14 R$+!$+%$+<@$+>$* $@$>13$3<@$4>$5:::$1!$2 R$+!$+<@$+>$* $@$3$4!$1!$2 R$+%$+<@$+>$* $@$>13$2<@$3>$4:::$1 R$+<@$+>$* $@$2$3!$1 handle user@host case (no %s) R$*<@>$* $@$1 local, or something # must be the $: field R$*@$+ $1 remove rightmost R$+%$+ $:$>6$1%$2 get a new @site R$+@$+ $>12$1<@$2>.XXXX redo, with brackets this time ##### Ruleset 20 -- No-op rewriting S20 ################################################### # Mailer Descriptions # ################################################### Mlocal, P=/bin/mail, F=rlsDFMmn, S=10, R=20, A=mail -d $u Mprog, P=/bin/sh, F=lsDFMe, S=10, R=20, A=sh -c $u Muucp, P=/usr/bin/uux, F=DFMhuU, S=11, R=12, M=100000, A=uux - -r -z -a$f -gB $h!rmail ($u) Msmtp, P=[IPC], F=mDFMueCX, S=17, R=20, A=IPC $h ##### Ruleset 17 -- Sender address rewriting for ethernet # if it is in uucp ! format, let's just leave it that way ok? S17 R$-!$+<@$+>$* $:$>12$1!$2<@$3>$4 if in ! format, rewrite @s to !s R$~w!$* $w!$1!$2 .. prepend our name R$=w!$* $@$w!$2 ... all done R$*<@$+>$* $@$1<@$2>$3 already ok R$*<@>$+ $@$1<@>$2 also ok R$+ $@$1<@>$w tack on our hostname ------------------------------ Date: 1 Jun 87 22:09:39 GMT From: richl%penguin.uss.tek.com@relay.cs.net (Rick Lindsley) Subject: Make your workstation a bit more secure One common problem is that as computers shrink, they are moving to desktops, and as they move to desktops, they become increasingly accessible. On my Sun, for instance, anybody who happens to walk by machine could halt it, bring it up single user, and do whatever they wanted to (which may not agree at all with what *I* wanted!). That is, until now. I offer this simple little program as a means of foiling these people. Chkpass will verify a password from within a shell script, which means you can now put something in /.profile to prompt for root's password. Everytime the system comes up (or goes down) single user, someone must now supply a root password. Or more accurately, just "a" password -- it need not be root's. I've made it root on mine, but if you want to make it something else, be my guest. Here's my /.profile, using chkpass from /local: stty erase ^H kill ^U PATH=/etc:/usr/etc:/usr/ucb:/bin:/usr/bin:/local export PATH TERM if [ `tty` = /dev/console ] then trap '' 2 3 ok=no while [ $ok = no -a -f /local/chkpass ] do /local/chkpass root case $? in 0) ok=yes;; 1) echo Sorry;; 2) echo Something\'s wrong with passwd; I\'ll allow you this time. ok=yes ;; 3) echo chkpass improperly invoked\; allowing root for now.... ok=yes;; *) echo unknown error status from chkpass\; allowing root for now ... ok=yes;; esac done trap 2 3 fi The sanity check for /dev/console requires that "tty" be somewhere on the root disk, which it normally isn't. For my purposes, I put a copy in /local with chkpass. You can do without this sanity check if you wish; it just might make people supply the password twice in some situations. Of course, one must be very careful when messing with /.profile, so for that reason the above code provides an "out" in many situations. See the man page below for a description of the exit codes. So far as I know, this method, though simple, is secure. Keyboard generated interrupts are ignored during the password prompting by both /.profile and chkpass. (/bin/sh can't ignore TSTP, but then if your single-user shell is /bin/sh, it won't be generated either, so ....) The beauty of it is you don't need to modify any existing utilities if you don't want to (though one may argue init is the program that should be doing this sort of thing.) Suggested modifications are welcome, but one thing I like about this program the way it is right now is it fits so nicely into shell scripts as a general purpose tool. Rick Lindsley richl@penguin.uss.tek.com ...!tektronix!penguin.uss.tek.com!richl .TH CHKPASS 1 "April 23, 1987" .SH NAME chkpass \- get a password from within a shell script .SH SYNOPSIS \fBchkpass\fP \fIuser\fR [prompt] .SH DESCRIPTION Chkpass prompts for a password on /dev/tty (via \fIgetpass(3)\fR) and verifies it with the entry for \fIuser\fR in /etc/passwd. .LP If the password field in /etc/passwd is null, then chkpass will not ask for a password, and is essentially a no-op. .LP The second argument, the prompt, is optional. If it contains white space or characters special to the shell, it will need to be quoted. .SH EXAMPLES .nf .in +0.5i % chkpass richl Password: % .sp 2 % chkpass richl "Gimme key:" Gimme key: % .fi .SH "EXIT CODES" An exit code of 0 indicates a correct password, or no password necessary. .sp An exit code of 1 indicates an incorrect password. .sp An exit code of 2, as well as a message to stderr, indicates the user was not listed in /etc/passwd. .sp An exit code of 3 indicates chkpass was invoked without the requisite \fIuser\fR argument, and provokes a usage message. .SH FILES /etc/passwd .SH "SEE ALSO" getpass(3) .SH AUTHOR Rick Lindsley, Unix Systems Support, Tektronix #include <stdio.h> #include <signal.h> #include <pwd.h> /* * chkpass -- prompt for a password and check that it is correct. * The first argument is required, and is the user name * to check against. The second is optional, and is the * prompt desired. Keyboard generated signals are ignored, * thus this can safely be used from shell scripts. */ main(argc,argv) int argc; char **argv; { register struct passwd *pw; register char *pass, *encrypted; char *getpass(), *crypt(); (void) signal (SIGINT,SIG_IGN); (void) signal (SIGQUIT,SIG_IGN); (void) signal (SIGTSTP, SIG_IGN); if (argc != 2 && argc != 3) { fprintf(stderr,"usage: chkpass username [prompt]\n"); exit(3); } if ((pw = getpwnam(argv[1])) == NULL) { fprintf(stderr,"chkpass: can't find user \"%s\" in /etc/passwd\n", argv[1]); exit(2); } if (pw->pw_passwd && strcmp(pw->pw_passwd, "")) { pass = getpass(argc == 2 ? "Password:" : argv[2]); encrypted = crypt(pass, pw->pw_passwd); if (strcmp(encrypted, pw->pw_passwd)) exit(1); else exit(0); } else exit(0); } ------------------------------ Date: Thu, 28 May 87 15:34:27 CDT From: hood@themis.rice.edu (Robert Hood) Subject: Sun workstations for the home? Suppose one wanted to have a workstation with a disk drive but not connected to an Ethernet (at home) that needed to share files with a remote network (at work). The machine at home would be connected to the network with a dedicated phone line (perhaps 9600 baud.) In particular, I want to be able to edit program source files, compile them, and then run the executable--both at work and at home. (There are 105000 *lines* of source code and the executable is over 3 megabytes.) Is there any software that would help manage this? Ideally, one could imagine some variant of NFS that allowed multiple copies of a file to exist on different servers (and kept them consistent). Failing that, is there some reasonably painless way to transmit updates to source files back and forth? (It would be nice to just transmit differences between files, but I might be willing to settle for entire file.) Has anybody else run into similar problems, and if so, what solutions have been used? Robert Hood hood@rice.edu ------------------------------ Date: Thu 28 May 87 14:01:55-PST From: SPRINGER@star1.stanford.edu (George Springer) Subject: 3rd party disk drives? Earlier this year we got a standalone Sun-3/160 with a Fujitsu 2322 130Mbyte SMD drive and a Xylogics 450 controller. We're running short of disk space faster than expected so we're in the market for another drive that could be hooked up to this controller without conflict with the 1st drive. As is usual in academic situations, money is tight. On the other hand, our expertise with this stuff is very limited - we don't want to have to write our own device drivers. So if you can suggest a drive model that we should investigate and/or a place to buy them that offer good prices, we would greatly appreciate it. Thank you for your help ------------------------------ Date: 29 May 87 12:42:38 GMT From: mrs@philabs.philips.com (Mark R. Simpson) Subject: Color Hard Copy Units? I am interested in any comments, suggestions, or recommendations about equipment for producing color hard copy from a color workstation. The color images will range from simple 4/8/16-color business plots (like those produced by DISSPLA or CricketGraph) up to realistic shaded images. However, the most common images will be from Solid Modelling/ Finite Element Analysis type programs on a color SUN 3/260 (8 bit planes). It would be nice if the color hard copy unit could be used by different types of workstations (we have SUNs, APOLLOs, an IRIS and a PIXAR). Furthermore, since all our computers are connected by a LAN, a unit that connects to a local ethernet(TCP/IP) would be a plus. Some specific units I have heard of are: 1) Shinko CHC-65 2) Tektronix 4692 3) Seiko CH-5300 4) Color Versatec If you have experience with any of these products or others you believe deserve mentioning, please e-mail me your comments and I will post a summary to the net. Possible considerations are: 1) cost 2) copy quality (e.g. number of colors, how closely colors match those displayed on the screen, etc.) 3) paper sizes it supports (A,A4, etc.) 4) how it operates. For example, do images need to be displayed to be copied? Is it connected directly to the RGB lines? 5) print mechanism. Is it a camera, an electrostatic plotter, ink jet plotter? Thanks in advance, Mark Simpson Philips Laboratories Briarcliff Manor, NY 10510 (914) 945-6163 P.S.: Recently, there was a similar request posted to the net, but I never saw the summary. ------------------------------ Date: 31 May 87 01:34:04 GMT From: ucsbcsl!bertrand@seismo.css.gov (Bertrand Meyer) Subject: Clean version of dbxtool? Our version of dbxtool on the Sun-3 has the unpleasant tendency to die without warning or possible recovery. Does anyone have a clean version? Bertrand Meyer, Interactive Software Engineering 270 Storke Road, Suite 7 - Tel (805) 685-1006 Goleta, CA 93117 ------------------------------ Date: Thu, 28 May 87 22:42:12 EDT From: schoff@nic.nyser.net (Martin Lee Schoffstall) Subject: sun's ie connected to a DECOM? I would like to attach my sun3/260 ethernet interface (ie) to a DEC DECOM. I tried to do this awhile ago for fun with little success, NOW I find it is necessary. I believe that the problem is the heartbeat. Has anyone solved this problem for the decom (by modifying the ie driver) or for any of the other ethernet in a broadband situations? marty schoffstall ------------------------------ Date: Sat, 30 May 87 0:18:20 EDT From: scs%lokkur%itivax.uucp@eecs.umich.edu (Steve Simmons) Subject: 3.0/3.2 .rhosts and 3.2 diag? We have noticed an interesting and frustrating bug in the treatment of the .rhosts files on Suns under 3.0 and 3.2. If there is trailing whitespace on the id mentioned, it is treated as part of the id and hence the id no longer works. Eg, the following line in /.rhosts on system "fred" "admin root" will let root@admin rlogin immediately, but the line "admin root " will refuse the connection. Also, we have been using CDC-9771 drives with our Sun-2 and Sun-3 systems for some time now, and had reasonable luck. When converting to 3.2, it seems like any time we touch the disk with 'diag' the disk shortly thereafter becomes corrupt. Sometimes labels are lost, sometimes boot blocks are destroyed, etc, etc. We have one major suspect in that some things changed between 3.0 and 3.2 'diag'. For one thing, 3.2 asks for some block size and interleave data which 3.0 did not. Does anyone out there know what defaults the 3.0 systems used, so we can put in the "hopefully" right answers while bringing up diag? If we diag and partition under 3.0, then load and setup with 3.2, things seem to go fine. Wierder and wierder... Many thanks, Steve Simmons ------------------------------ Date: Sat, 30 May 87 19:20:13 PDT From: weitek!sci!daver@sun.com (Dave Rickel) Subject: Function key programming? This is probably old hat, but does someone know how to program the function keys (F1-F9)? I want to turn off F1--it's too close to the escape key. decwrl!sci!daver ------------------------------ End of SUN-Spots Digest ***********************