« Harmonic Hot Air, Part 2 | Main | Ph’nglui mglw’nafh Cthulhu R’lyeh wgah’nagl fhtagn! »

November 08, 2005



Sigh. Yes. Certainly CS departments could take a first step in how they teach computer science to other fields. For example, why do non-major courses teach C++? What does a physicist need to know about C++? Teach python. Teach whatever scripting language you like. Teach something which is useful to them but through which you can demonstrate CS ideas.

And I don't care if you use emacs or vi.

Daniel Lemire

Gosh. You guys work at great schools...

> emacs is better than vi

Most of my colleagues seem to use neither. At least, I keep getting word documents in my inbox that I have to painfully open using OpenOffice.

> Eclipse better than IDEA

You guys debate this sort of stuff?


>For example, why do non-major courses teach C++?
I notive the same thing at my school. It's pretty weird. Is it somehow considered "inferior" to Java?

>In the age of information nearly no discipline at a large scale >requires computer science, certainly not programming as we teach >it.

Indeed, I doubt the emacs vs vi debate holds any interest for the common man.


"You guys debate this sort of stuff?"

No, we're more of a Windows-versus-open-source crowd (which is, admittedly, just as bad).

"Why do non-major courses teach C++?"

Because there's no "Numerical Recipes in Python". (shudder)


>Because there's no "Numerical Recipes in Python". (shudder)
Any other reason would have worked. This one doesn't. How may of us really need to rewrite BLAS, LAPACK, etc routines? Most of us are happy to deal with python wrappers (scipy, numeric, etc) or matlab/octave most of the time.

RE:We're doomed: You've probably read Eugene's entry titled: "Day 4: Mary Beth Rosson on the End of Users" at http://www.cs.uni.edu/~wallingf/blog/archives/monthly/2005-10.html#e2005-10-19T18_14_36.htm
about user developpers. Do you think user developpers might save us?


"How may of us really need to rewrite BLAS, LAPACK, etc routines?"

None. But how many students are taught to write them anyway?

Michael Stiber

When's the last time you heard of a CS department stepping up to the plate to create a service course that really satisfies the needs of those other departments? Math and the sciences are fundamental to other fields, and so their introductory courses have wide currency. CS departments approach their non-majors courses the same way. But CS is not fundamental to those other fields in the same way. But our intro courses for non-majors are designed like "CS for slower students", rather than "CS for students who want to use it as a tool in their own fields". Rather than "ratcheting down" a freshman course for CS majors and selling it as a non-majors course, we need to do some ground-up design for, say, Sophomores or above from other fields. The onus is on us to create something compelling enough that those other departments would want to require their students to take it (and wouldn't feel that they could do better).

Oh, and using a language like MATLAB sounds like a good way to break out of the mental box in this respect. It would also likely be closer to the style and level of programming science & math students would actually do.


As one who still believes that real programmers use Fortran, I was Googling around for something on this subject and ran across this good article:
by a physics professor who wishes his department required their students to learn programming from the CS department.

According to the article, two of the significant changes that have occurred in the past 20 years are (1) the decreasing per cent of science students who study programming in high school and (2) that desktop computers no longer come with interpreters and compilers. These and other changes mean that today's science students need programming courses more than previous science students did.


As a biologist, I can say that it would be supremely useful to have CS people teaching us basic UNIX and PERL. These would be highly applicable to our research, and is likely to become more and more common - even necessary. That is, unless someone goes and invents yet another programming language...


> Because there's no "Numerical Recipes in Python". (shudder)

Why the shudder? Python's widely used in the sciences, mostly for bioinformatics and SciPy (http://numeric.scipy.org/) offers fast numerical methods too.


The shudder was for "Numerical Recipes"—ie, the specific series of books with that title—not for python.

Ernie, Tell us about Paris.

The comments to this entry are closed.