[Beowulf] Teaching Scientific Computation (looking for the perfect text)

Robert G. Brown rgb at phy.duke.edu
Tue Nov 20 15:01:31 PST 2007


On Tue, 20 Nov 2007, Peter St. John wrote:

> they said they didn't know why. It turns out that the automaton
> produces C which is morphed to look like Fortran, so for example this:
>
> 100 WRITE (6, 200, X)
> 200 FORMAT(...)
>
> turns into
>
> fortranlikeWRITE(output_device_6, "LineNumber200", X)
> LineNumber200: fortranlikeFORMAT(...)
>
> instead of
>
> fprintf(stdout, "%d", X);
>
> so d'uh, no wonder that their clients maintain this code in fortran,
> and not in C.  So there are some fortran progammers with jobs around
> here, good :-)

The evils of f2c.  Been there, used that, decided that it was FAR BETTER
to just start over, read the goddamn fortran sources, and convert them
in to nice fresh C.  My original port of diehard went just like this, as
the "C source" was of this especially pernicious sort, and the I/O was,
well, fortran file based I/O -- does it get any uglier?

> I don't draw any conclusions about the futures of any languages:
> really I think it won't matter much as compilers get smarter and we'll
> express ourselves any way we like. Also, sorry for the fudgy fortran,
> I'm very rusty.

Resisting... temptation... to... continue... rant...

Compilers?  Smart?

It would take compilers that read your mind to replace a single human
capable of expressing PRECISELY what the computer is supposed to do,
telling the computer to do precisely the right thing.

Remember, "smartness" is a process of Bayesian reasoning, which in turn
is based on a map of joint, multivariate probabilities.  PLC used to do
its damnedest to compile any old thing you threw at it by guessing what
you meant.  People used to have (expensive) fun throwing decks of cards
picked up at random off of the floor and feeding them to it.  They'd
compile.  God knows to what, but they'd compile.

I want a stupid language that lets me be the smart one in the
computer-programmer relationship.  I want the computer to do exactly the
one thing that my language statements unambiguously tell it to do.  If
those are the wrong thing, well, I want it do to the wrong thing
perfectly and bomb with the greatest possible grace.  Then I can fix my
code and have it be the right thing perfectly.

How do you tell your boss "The compiler misunderstood me..."?

    rgb

>
> Peter
>
>
>
> ...
>> I have a staff of 5. Of these 3 are conversant in Fortran. Our ages are
>> 52, 43, and 40. We deal with fortran on a daily basis to support our HPC
>> users in Physics, Chemistry, Mechanical Engineering, Chemical
>> Engineering, and occasionally even statistics. Supporting these users
>> represents over $20 Million of direct funding to the University.
>>
>>
>> I remember a time in the mid 1980's when I questioned the value of
>> Fortran. But it's still here, it's still used, and we still need people
>> who understand it to answer the next generation of questions in the
>> basic sciences.
>>
>>
>> Mike Davis
>>
>>
>> _______________________________________________
>> Beowulf mailing list, Beowulf at beowulf.org
>> To change your subscription (digest mode or unsubscribe) visit http://www.beowulf.org/mailman/listinfo/beowulf
>>
> _______________________________________________
> Beowulf mailing list, Beowulf at beowulf.org
> To change your subscription (digest mode or unsubscribe) visit http://www.beowulf.org/mailman/listinfo/beowulf
>

-- 
Robert G. Brown
Duke University Dept. of Physics, Box 90305
Durham, N.C. 27708-0305
Phone(cell): 1-919-280-8443
Web: http://www.phy.duke.edu/~rgb
Lulu Bookstore: http://stores.lulu.com/store.php?fAcctID=877977



More information about the Beowulf mailing list