NetNews Usenet Archive 1992 #20/NN_1992_20.iso/spool/comp/lang/perl/5802

Internet Message Format | 1992-09-09 | 5.3 KB

Path: sparky!uunet!eiffel!eiffel.com
From: [e-mail] (Raphael Manfredi)
Newsgroups: comp.lang.perl
Subject: Re: How to convice awk/sed users to use Perl ?
Summary: Apology for perl
Message-ID: <[e-mail]>
Date: 9 Sep 92 18:07:35 GMT
References: <[e-mail]>
Sender: [e-mail]
Organization: Interactive Software Engineering, Santa Barbara CA
Lines: 88

Quoting [e-mail] from comp.lang.perl:
>As a Perl beginner, I don't know how to answer the following question raised
>by my colleagues:
> I use ksh/awk/sed very well, why bother to spend a lot of time to learn Perl ?
 What's the real advantages of perl over ksh/awk/sed ?
>
>I will appreciate any help.

There are multiple facets in your question, upon which I shall try to build
my argumentation.

To start with, perl has been cleverly designed to integrate most of the
attractive features from the C, shell, sed and awk into one single language.
This means you only have to deal with one (hopefully consistent) program
instead of three of four dissimilar tools, each of them having its own
conventions and idioms. You don't have to "think C" and then later "think sed"
or "think awk", you simply "think perl" and are able to easily put your
ideas in a concrete form. To mention but the most immediate advantages perl
brings you: reusability, ease of maintenance, uniformity, flexibility.

Perl brings you reusability, not the "true" one object-oriented languages like
Eiffel would bring you, but still you may encapsulate functions in libraries
or software components and then share them across your programs. I have to
concede that the Korn shell brings you that too, since it has functions and
files can be "dotted" the same way perl "requires" them. However, sed and awk
lack this possibility.

Perl brings you ease of maintenance. This is a relative notion of course, since
maintaining a huge perl program may become tricky (for instance, perl does
not have static scoping -yet-), but I claim it is simpler to maintain a single
perl program than a set of ksh/awk/sed scripts performing the same operation.
And it is much easier to abstract things in perl than it is with sed or awk,
although perl does not currently support recursive data structures very well.
And perl comes with an interactive debugger, making it "easy" to track down
possible problems. In particular, you get step by step execution.

Perl brings you uniformity. Since it is one language, the constructs it is
using are integrating quite well, as opposed to the unpleasant discrepancy
you get with ksh/sed/awk. Furthermore, you are facing the famous wizard's
dilemma of choosing the right tool for each part of your software. I don't
think someone would dare to imagine doing pattern matching in ksh or process
control in sed, but if you want to do both, then you need at least two
cooperating parts in your program, one in ksh and one in sed. Communication
issues left aside, you are faced with a single program using two different
tools, with different conventions, whereas perl would integrate those two
parts elegantly.

Perl brings you flexibility. There is generally more than one way to do things
in perl (i.e. the language is not minimal) but this lets you use whatever
scheme or idiom you prefer. If you want to impress academics, one-liners are
usually good, whereas if you strive to design your program elegantly so as to
please the software engineer people then perl's name space protection via
packages or dynamic scoping brings you some of the advantages of the object
oriented world, which, I have been told, is nothing but eyewash :-)

There are some other advantages in using perl, but they have never been the
decisive factor for me. Nonetheless, they are considered important in the long
run if you happen to develop a lot of programs and make an heavy use of perl
scripts and programs (depending on your terminology, but it's six of one and
half a dozen of the other).

Perl is fast. In fact, perl is optimized for pattern matching (some people say
it comes from its name, but I don't believe them :-), and depending on your
application, a perl script can be faster than a corresponding awk script.

Perl is interpreted. But that is also true for all the alternatives you
mentioned, therefore this is not a true advantage. At least there is no
big conceptual jump between ksh/awk/sed and perl (since compiled programs
often have some restrictions which are happily ignored by interpreters).

Perl comes with the source code. This may or may not be an advantage.

Perl is not only a UNIX tool. Apparently perl has been ported to other
operating systems and platforms like MS-DOG or the Amiga.

Now if your colleagues have something to say in the defense of ksh/sed/awk,
I would be glad to hear it. Note that I occasionally use sh scripts, or
mixture of sh/sed/awk, but usually I have a good reason for doing so ;-).
For instance, the Configure script which comes with perl has to be written
in shell + sed + awk. Agreed, it could be written in perl and made more
portable, but then you have a serious bootstrapping problem.

In the hope this will somewhat address your concerns...
--
Raphael Manfredi <[e-mail]>
Interactive Software Engineering Inc.
270 Storke Road, Suite #7                      / Tel [phone] \
Goleta, California 93117, USA                  \ Fax [phone] /