[Physltest] [Phys-L] Re: Mac vs. PC (no flames please!)

[John Denker <jsd@AV8N.COM>]

John SOHL wrote:

> We are discussing what type of computers to install in a modest
> computer teaching/research room. The obvious question is Mac vs. PC.

At the next level of detail, the question is threefold:
   Mac/MacOS
   PC/msWindows
   PC/Linux

> Nature of the room as envisioned:
> 1. Five to ten computers.
> 2. First priority goes to undergrad research (mostly astrophysics).
> 3. Second priority goes to classes such as computational physics ( C++,
> FORTRAN) and data acquisition (LabVIEW mostly).
> 4. Third priority goes to physics majors needing to do homework, email,
> etc.

That makes sense.

> Here are the questions that I need data for:
> 1. What platform is used the most (percentages??) in industry?

Windows (>85%) ... counting all industries, including lots of
not-very-technical industries.

> 2. What platform is used the most (percentages??) in grad school by
> theorists?

   PC/Linux

> 3. What platform is used the most (percentages??) in grad school by
> experimentalists?

   PC/Linux

> I tried a few searches with Google but clearly didn't have the search
> terms that I needed to get useful hits. Seems to me that someone on this
> list has likely faced the same question as to what platform they wanted
> to train their physics majors on. Opinions may be interesting but what I
> really need is some solid estimates of percentage of use.
>
> One faculty member is saying that, "EVERYONE in astrophysics uses
> Macs." Another faculty member is saying, "NO ONE in industry uses PCs."

The second one is completely bogus.  I wish it were true, but
it's nowhere near true.

Here are some desiderata:

  -- HW cost:  PC hardware gives more bang for the buck than Mac.

  -- SW cost:  Linux gives infinitely more bang for the buck
   than Windows.  It's free.

  -- Maintenance cost:  You might have to pay more per hour to
   get good Linux sysadmins, because they are scarcer ... but
   you don't need as many hours, because the software is better
   behaved.  Lots better.

  -- Reliability:  No contest.

  -- Security:  No contest.

  -- Life-cycle cost:  I'm composing this email on a laptop
   that was mfgd in 1998.  It's running a Linux release
   that is somewhere between fully up-to-date and bleeding
   edge (Debian Sarge) ... and it runs just fine.  Also
   programs that I wrote 20 years ago continue to run fine.
   In contrast, you wouldn't want to mix old applications
   and/or old hardware with new ms operating systems.

   This really affects your budget, because it changes how
   often you need to run around and replace hardware.

  -- Compatibility:  Almost every scientific application I've
   ever heard of runs fine under Linux.  This includes LabView.
   Students who are addicted to word-for-windows can use a
   work-alike thing under Linux.  There are a lot of popular
   computer games that won't run ... but that shouldn't be a
   consideration when buying classroom machines.

  -- About the only good thing about MacOS and Windows is
   that they both run Eudora.  I like Eudora, and I don't
   know of a comparably nice MUA for Linux.  This is a
   big deal, since I do a whooole lot of email, but it's
   nowhere big enough to make me switch operating systems.

-- Another goodish thing (in the silver-lining sense) is that
   you can install Cygwin under msWindows, which gives you
   a surprisingly decent unix-like programming environment:
   You can run emacs and G++ under Cygwin under msWindows.
   In particular, I've written programs that crash the machine
   (resource exhaustion) if compiled and run using microsoft
   C, but the same program (verbatim) works fine under Cygwin
   on the same machine.  It also works fine under Linux.  I
   collaborate with some folks at a fascist institution
   where everybody has to use PC/Windows ... so we just
   install Cygwin when nobody's looking.

   In general, therefore, even though you might think
   there's an ineffably large amount of software that
   runs under Linux but won't run under msWindows,
   that's not true, because of Cygwin.  But it remains
   a lot easier and cheaper to just run it under Linux.

True fact:  Once upon a time, I got caught up in a lab-wide
push to convert everything and everybody to PC/msWindows.
I converted hundreds of thousands of lines of code and
prose to the new platform.  I lived with it for two years,
so my understanding of the issues is not just theoretical.
Eventually I couldn't stand it any more and converted
everything back.

You've heard there's no zealot like a convert?  Hah!
A convert is nothing compared to a re-converted apostate.
_______________________________________________
Phys-L mailing list
Phys-L@electron.physics.buffalo.edu
https://www.physics.buffalo.edu/mailman/listinfo/phys-l