From: Christopher Browne
Subject: Re: Blocks Help When Writing Code? How? Re; 1+
Date: 
Message-ID: <qlos3.3077$gO1.120927@news2.giganews.com>
On 11 Aug 1999 14:47:08 GMT, Julian V. Noble
<···@node15.unix.Virginia.EDU> wrote: 
>····@ricochet.net writes:
>	[ mucho deleted ]
> 
>> I agree with Chuck that one of the most common mistakes people make
>> is that they want to use files for *everything*.  Many people use
>> files at runtime when they could use static memory, dynamically
>> allocated memory, or blocks.
>> 
>> I think when you come from an enviroment where you are encouraged
>> to think of everything as a file people will have a lot of files
>> that they really don't need to have.

... Which also parallels the popular misconception that Lisp is "just
about lists and recursion," when it has perfectly good array/vector
data structures as well as looping constructs.

And, in similar fashion, the APL world has some tendancy to treat
"everything as a vector," which sometimes reflects a useful
abstraction, but which sometimes results in trying Too Hard to Avoid
Loops.

In Perl, the "hammer" is that of associative arrays, and the incentive
is to turn all the "nails" into data congruent with that view.

>> I recall an excellent example in c.l.f a few months ago.  Someone
>> wanted to use DBase III to configure the control parameters for
>> his small embedded application.  

... Which could be either good or bad ...

Trying to push everything into the table/field abstraction can be bad,
when it's not appropriate.

In the same manner, directories/files is remarkably useful, but not
*always* so.  Block-oriented storage is a useful abstraction that,
particularly for "things that look like databases," can be rather a
lot faster and more robust than files.

>I really think this deserves to be in Programming Pearls. Pity
>Jon Bentley is no longer writing such a column for Communications
>of the ACM.

Indeed.  He'd doubtless be able to do a good column or five on the
somewhat subject that might be characterized as "Using The Right
Data/Control Structure For The Job."

-- 
"I've discovered that P=NP, but the proof is too long to fit within the
confines of this signature..."
········@ntlug.org- <http://www.hex.net/~cbbrowne/lsf.html>