🚀 go-pugleaf

In article <1cseutINNk30@fido.asd.sgi.com> shankar@sgi.com (Shankar Unni) writes:
>Herman Rubin (hrubin@pop.stat.purdue.edu) wrote:

>> I would like to see how you anti-gotos would write code of the following
>> type so that either a human being could understand it

..... schematic code deleted

>If anyone around here (read: "real world" (tm)) wrote code like that,
>he or she would be fired pronto, and the code sent into the bit
>bucket.

>Dr. Rubin, this sort of code may be OK for one of your throwaway
>research projects, but around these parts, people tend to move around
>from job to job, and I pity the poor slob who had to maintain crap
>like this.

>Sometimes, giving up 5% of the cycles in order to get clean, readable
>code with a reasonably clear flow of control is a perfectly acceptable
>tradeoff.

>If the loops get nested too deeply to be able to break out of neatly,
>it is time to restructure the code (break it up, break off some code
>into functions, etc.).

>After all, next year, all the CPUs will be 25% faster, whereas the
>code may have to be maintained, and worse, modified, for years and
>years to come.

-----------

I would not want to be in the position of defending Dr. Rubin's
rather exotic theories on programming but the above is excessively
provincial.

Primus -- Not everyone in the world is a vendor of hardware or software;
some people actually use computers to compute with.  The "real world"
of people writing commercial software for sale is not the whole real
world of computing.

Secundus -- Some algorithms are complicated.  Rubin needs a fast
implementation of an intriniscally messy algorithm.  Babbling about
clean, readable code [and breaking off code into functions!] is beside
the point.  Yes, production software for delivery wants to be clean and
highly maintainable.  Giving up 50% of the cycles [and 50% is a better
rule of thumb than 5%] is well worth it, in general.  But it is really
a perversion of engineering practice to insist that guidelines that are
appropriate in a particular domain are therefore universal for all
domains -- of course, Rubin is guilty of the same thing.

It is beside the point that tomorrows machines will be faster; Rubin
needs all the speed he can get today.  When tomorrow's machine comes
along he will still need all the speed he can get.

Now he has presented a task which he claims cannot be efficiently
rewritten without gotos.  He may be right -- case structures are
usually less efficient than the equivalent goto formulations because
of the checking and setup that is done.

More generally, he is making the point that there are people who have
to milk the machine for every last cycle, at least in critical code.
Standard languages are not designed for such people.  Nor should they
be; the number of people who need that kind of optimization and can
actually make use of it is small compared to the number of people who
will make a mess of their code while trying to exploit optimization.
-- 
Richard Harter: SMDS Inc.  Net address: rh@smds.com Phone: 508-369-7398 
US Mail: SMDS Inc., PO Box 555, Concord MA 01742.    Fax: 508-369-8272
In the fields of Hell where the grass grows high
Are the graves of dreams allowed to die.