RE: ash vs. bash
From: Steve Lamb [mailto:email@example.com]
On Wed, 21 Jul 1999 19:09:48 -0400, Raul Miller wrote:
>(1) it's easier to learn a language that you can use interactively than
>one that only works in some "file at a time" mode.
This is an opinion. I simply counter with mine. I find that one that
goes through a file is easier to learn than one that does not. My
with perl (which I know) versus sh/ash/tcsh/bash/zsh (which I do not know)
well as MUSH coding on a MUSH versus MUSH coding in a file and preprocessing
into something usable.
IE, just because something is interactive doesn't mean it is easier to
>(2) it's often convenient to write a short script at the command line to
>accomplish some one-off task.
Again, an opinion. I really see no difference than writing a quick
at the prompt versus writing a quick script. Well, I do. when I screw up
the prompt I have to really look to see what the problem is, maybe even
retyping large portions because of a lack of decent editing.
Steve C. Lamb | I'm your priest, I'm your shrink, I'm your
Steve's argument is from the proscriptivist "less is more" stance. Along
the lines of, "the world will be more rational and easier to cope with if we
first eliminated all the unwise methodologies" (such as frittering away a
lot of time trying to perfect a complex script interactively).
Anarchy is indeed an exceptionally pure form of democracy. However, IMHO
there's a lot that can be said here before the skies are obliterated by the
dark cloud of contrasting opinions.
There is a version of the Laffer fallacy at work here. At one end of the
spectrum we look at how novices work and conclude-if we are so inclined-that
their progress would benefit from a more structured approach (i.e. file
based). And then we look at the experts who know exactly what they need to
do and how to do it and we can again conclude that interactivity is a frill.
And the frosting on the jello is to draw a smoothly curved "Laffer" in
between these endpoints and thereby pronounce "case closed".
Unfortunately, the C++ standardization committee fell victim to these same
urges. There are instances where a template structure is valid in a class
if that class is in global scope, but not if that class is itself nested
within a parent class. These rules actually increase the burden of
authoring a compiler, since once you are finished with the general case you
still have to go in a whack away the prohibitions.
The "fifth business" is the point in the middle of the curve where the
"Alice in Wonderland" scenario applies and the system is simply behaving in
incoherent and contradictory ways. In these situations every duality is a
blessing and every proscription is a curse. The more things you can try the
sooner your feet find solid ground again.
Only in the clean world of computer science can we dare to reach conclusions
(aka Pascal/Java) about what features of the language/environment are
meritorious. It's this middle ground where the Laffer curve does an
unexpected loop that you learn to value every degree of freedom which the
system affords and which IMHO the "clean" approaches unfairly discount.
(PS: Pardon my "Outlook" - any hints on how to get this bloated creature to
belch out mail in a "rest-of-the-world compatible" format are welcome.)