re: Compatibility

[turnbull@example.com (Stephen J. Turnbull)]

First, let me say that I agree with Ted, it's nice to buy well-tested,
tried and true hardware and put software tested with that hardware on
it.  What do you do when you inherit a poorly integrated system?  Or
you want something *really* high performance, bleeding-edge, but can't
afford SGI?  You take your chances, and you make it work.

I *am* a grown-up.  I'm not complaining about NT crashing my hard
drive.  I was backed up.  What I want to do is (1) to compare the MS
attitude toward these problems with the Linux attitude (MS fixes
problems that cost them lots of customers, ie, money, Linux gives you
source and you fix your problem if nobody else will do it for you---
but usually they will), and (2) to argue that Linux can be quite
transparent (although it does require much more effort to set up than
Windows).  Oh, I shouldn't forget (3): flame on a bit.

Basically (2) gets no space in the rest of this post.  :-)
I plan to put together the WadaLab fonts configuration script
tomorrow, though, so you'll see my contribution to transparency
shortly.  :-) :-)

>>>>> Ted writes:

   MS [snip] may be big, but why is big bad?  It's not.  It's good to be
   big.  It's good to make money and return value to your
   shareholders.  Don't forget that.

Stephen John Turnbull
Ph.D., Economics, Stanford University, 1985.
"You don't need to teach your grandmother how to suck eggs."

You didn't read what I wrote, did you?  Power can be misused.  My
belief is that Microsoft uses its market power to ignore problems with
its products.  This is certainly true with respect to Win 3.1 and
Win95's DPMI.  (Not the 4GB security hole, this is a different one
that causes UAEs.)  That is surely returning value to MS shareholders,
otherwise they wouldn't do it.  That doesn't make it a good thing.

I would prefer that MS had less power.  If nothing else, they might
fix their DPMI and users who can't believe that MS would sell and not
fix buggy software would stop harassing my friend Charles Sandmann
(author of the DPMI facilities of DJGPP).  (He's sent MS a description
of the problem and why he's sure it's an MS bug; they didn't reply
when he suggested a fix, and they said nothing constructive to the
original bug report.)

It's very clear to me that MS is returning value to their shareholders
*and to their customers*.  I think MS Office is a wonderful thing ... 
for other people.  Ditto Win 95.  But I know too many problems with MS
software and with the whole environment that affect me personally to
be able to return to it comfortably.

I wrote: fdisk says "Hmmm... this disk is over the 1024 cylinder
   limit, you could have problems."

Ted replies: So don't use fdisk, I don't.

I *want* to know when my software has problems.  Besides, the fdisk
warning is not about fdisk, fdisk doesn't care.  There are other parts
of the system that *do* care and *have* trashed disks.  That's what
fdisk is warning about.

   Now in my opinion, you are not making well informed buying decisions.  We 
   have an Epson Esper printer, but no one will admit to having bought or 
   recommended it.  We all rely on proven industry standard HP and Canon 
   printers.  IMHO, Epson printers are crap.

Could very well be.  I didn't make that decision.  I would have bought
HP.  In fact, the Novell LAN wasn't my idea either.  I just have to
deal with the problems I've inherited, and the vendors are just
pointing fingers at each other.

   I won't comment on the AMI and "logic sisters" stuff, 

Smart idea.

   but it is highly possible that you could  have made a wiser buying decision.

Geez, you hadda open your mouth again, didn't you?

   PS: get a subscription to Byte or PC Magazine, and stop reading the subway 
   Epson ads with pretty girls.

I don't know any pretty girls who want to read subway Epson ads with
me.  I'd love to be introduced.  But hurry up: my fiancee will make me
stop doing *anything* with other pretty girls soon enough, anyway.

Byte is useful.  PC Magazine I find a little too specialized.  (Too
much emphasis on Windows software reviews.)  Computer Shopper helps a
lot.  comp.os.linux.* is pretty voluminous....  ASCII and DOS/V
Magazine are junk as far as I can tell (I've only really read one
issue of each, Japanese is too difficult to take a larger sample.)
CACM is fun, but doesn't have much direct relevance to Linux ;-)
SIGACT News has *no* relevance to hardware choice ;-) ;-)

   Subscribe to newsgroups where everyone in the real IT world can
   give you real world feedback.  Believe me, they won't point you in
   the buying direction's you've been heading towards...

The only problems I ever had with that board are (a) NT trashed my
disk, (b) I use an old revision of the BusLogic driver or an AHA1542
driver under Linux, and (c) the external SCSI port is a SCSI-2 50-pin
mini-D, and the vendor that the University in its infinite wisdom
chose to buy my HP scanner from has *not yet after 14 months*
delivered the necessary optional cable.

On the plus side, that 50MHz '486DX box regularly got transfer rates
of 400kBytes/sec from the local FTP server, 70kBytes/sec from
Tohoku-dai's FTP server, and I once recorded 1700 HTTPd contacts,
about 85% of them CGI requests, in a 100-minute period.  I think
that's pretty good performance for an EISA SCSI controller that cost
only $100 more than the street value of its cache....  Sure beats the
hell out of my new Gateway Pentium 120MHz PCI, or even a rather
expensive DECserver (although that box has only a 90MHz P5)---they
can't handle the disk I/O.  (The ethernet is an EISA 3Com 579, but I
don't think that makes that much difference, even over a NE2000.)

Seriously, I got very lucky on that system.  I picked it for all the
right reasons and paid far too little attention to the dangers.  But
that more or less randomly picked old hardware beats new hardware with
much more powerful specs, systems integrated by professionals.  That
doesn't speak well for conventional wisdom or the skills of systems
integrators.

-- 
                            Stephen J. Turnbull
Institute of Policy and Planning Sciences                    Yaseppochi-Gumi
University of Tsukuba                      http://turnbull.sk.tsukuba.ac.jp/
Tennodai 1-1-1, Tsukuba, 305 JAPAN                 turnbull@example.com