🚀 go-pugleaf

In article <1992Apr3.190000.23149@jato.jpl.nasa.gov>, vsnyder@math.jpl.nasa.gov (Van Snyder) writes:
>
>Sure, my PC crashes more frequently than my Unix box, but when it does, it
>only takes 15 seconds to reboot, not 15 minutes.

I'd gladly trade a slower reboot for higher reliability.  Also, I
think you're exaggerating a bit...most PC's take at least a couple
minutes to fully boot, and most modern UNIX workstations only take a
few minutes to boot.  E.g., a DECstation 5000/200 w/several GB of
disk rebooted in seven minutes, with full fsck's.  And lot's of
systems skip fsck's after clean shutdowns.

>I'm willing to put up with
>that to have a REALLY GOOD editor (brief or sage or ...) that I can afford
>($250 or less),

GNU Emacs, free.

>a REALLY GOOD word processor (WordPerfect 5.1 or Word or ...)
>that I can afford (~$500),

Share a FrameMaker license with a couple friends, or buy
IslandWrite/Draw/Paint for $1000.

>a REALLY GOOD Fortran development system with a
>REALLY FAST compiler and REALLY GOOD debugger that knows Fortran, not just C
>(Lahey),

I have a hunch such products are available, though I have no use for
them myself.

>a REALLY GOOD spread sheet, data base, ...

Wingz, eXclaim, etc... Again, don't know about DBMS's, but they're out
there.

>The problem with Unix applications is that those one can afford are really
>sleazy (e.g. vi -- one level undo? edit one file at a time? modeful interface?
>get real!)

I don't like it either, but it *is* powerful, robust, and cheap.

>or really sloppy (e.g. emacs takes 37 MB of disk space, brief fits
>in one MB),

DEC 5000/200, Ultrix 4.2:
-rwxr-xr-x  1 root      1574260 Sep  4  1991 /usr/local/bin/emacs

Sun SPARCstation ELC, SunOS 4.1.2:
-rwxr-xr-x  1 root      1593344 Jul 18  1991 /usr/local/bin/emacs

IBM RS6000/320, AIX 3.1:
-rwxr-xr-x   1 root     system   2737398 Jul 18 1991  /usr/local/bin/emacs

These all have X and dumb tty support.

With complete documentation (300+ pages printed), binaries, and LISP
support files including the CALC package, it's 9500 KB.  This seems
reasonable to me, considering the functionality provided.

>while the good ones are really expensive (compare Frame or
>Interleaf

Agreed, they're expensive.  But depending upon your need, you might be
able to share a single license with two or three others, making it
much more economical.

>or Island Write or Unix WordPerfect to Wordperfect or Word on a PC).

Yes, there's a gap, but it's closing.  Consider the size of the PC sw
market vs. the UNIX sw market.

>Maybe when YOUR workstation breaks some network related thing your Unix
>tells you more obviously what is going on.  I recently had just the
>opposite experience:  (1)  The sendmail.cf that was delivered with our
>system carefully stripped off the internet domain name.

Pilot error.  You can't expect the default configuration to suit all
potential users.

>(2) since it
>was sendmail.cf, which is the only thing I've ever seen that's worse than
>OS\360 JCL, I just got one from a colleague, and it seemed to work.  One
>doesn't look inside of sendmail.cf unless absolutely necessary, and one
>has almost no hope of changing it in such a way as to improve it.

UNIX Myth #23: sendmail.cf is incomprehensible

In fact, it's not that bad if you take the time to learn it.  If you
don't, then don't complain about it.

>(3) A gateway two hops away broke.

So?  Unless you have a really screwed-up sendmail configuration, it
should have attempted to send for three days or so before bouncing.

>(4) Nothing in my system told me that
>gateway was broken, since there was another gateway that gave me access
>to the same places.

Huh?  It gave *you* access to the same places but wouldn't route mail
to them?  Something is seriously broken here.

>(5) After a while (four weeks), I concluded that
>outgoing mail was disappearing down a black hole, even though mail was
>still arriving.

You can test incoming mail by indirectly addressing it yourself (or
someone else local) through an external host, eg: foo%mysite@yoursite.

>(6) After another two weeks I discovered that outgoing
>mail was being sent to the broken gateway no matter what the address --

Sendmail.cf, no doubt.  You really should learn how to read at least
the most important things in there.  Or hire someone who knows what
they're doing to set up your system.

>I stumbled onto /usr/spool/mqueue/syslog -- Isn't that an obvious place
>to look?

Absolutely.  You do know about syslogd and syslog.conf, don't you?

>After all, that's where all the other log files for all the
>other tools go, right?

Wrong.  It often makes sense to store log files with the software they
belong to, rather than dump them all into one "log" directory.  Of
course, most reasonable UNIX software can be made to exactly that, if
you know what you're doing.  If you *don't* know what you're doing,
and you aren't willing to learn, then don't complain that the default
configurations don't suit please you.

>I posted questions on the internet, and didn't get even one suggestion --
>apparently, nobody knew even where to start.

Or you posted to the wrong group(s), or you asked a FAQ, or you had an
attitude or...

>I don't have anything on my PC's except telnet, rlogin, ftp and nfs,
>so I don't know whether network related problems are bigger on PC's
>or smaller, but the above experience convinced me that Unix network
>problems can get arbitrarily bad.

A little education goes a long way...

--
Dave Sill (de5@ornl.gov)	         "Nobody likes nuclear bombs.
Martin Marietta Energy Systems            We're trying to get rid of them."
Workstation Support                                  -DOE Sec'y James Watkins