Re: Payware ??? (Dwight VandenBerghe)
16 Dec 1997 23:41:24 -0500

          From comp.compilers

Related articles
Payware ??? (Paul Mann) (1997-12-16)
Re: Payware ??? (1997-12-16)
Re: Payware ??? (1997-12-17)
Re: Payware ??? (1997-12-19)
Re: Payware ??? (Laurent Guerby) (1997-12-23)
| List of all articles for this month |

From: (Dwight VandenBerghe)
Newsgroups: comp.compilers
Date: 16 Dec 1997 23:41:24 -0500
Organization: All USENET --
References: 97-12-133
Keywords: practice

"Paul Mann" <> wrote:

>I'd like to know how many software developers are using a free
>compiler for their work.

I use VC++ (not free), Thompson Awk (not free), gcc, and perl (free).

>Then I'd like to know how good it would have been if it were free.

Well, Internet Explorer is from the same company, and it's free, and I
think it's pretty good.

>My toolkit is not free, because I'm not a professor at a university,
>wish I were but ...

Your toolkit is not free, not because you're not a professor at a
University, but because you have decided to charge money for it.
Richard Stallman (gcc) is not a professor, nor is Larry Wall (perl).
Nor are the authors of python, pccts, etc.

>If I were to give it away for free, I would be announcing version 1.0
>instead of 5.0.

Perl is currently at release 5.004.

>I've thought about giving it away for free. Any ideas, I'm listening

Keep selling it. I also use Yacc++, an absolutely excellent parser
generator that costs $995. It's worth every penny. The manual for
yacc++ is the best I have ever read for any software product, period.
And the thing works. If you can make a tool that is very much better
than what is available for free, why not profit from it?

The one big problem with payware compiler tools is that you will never
make any real money at it. Your freeware competition is pretty strong
(see pccts) and the good freeware tools tend to attract many strong
and fervent users who help each other, fix bugs, offer suggestions,
and so forth. Like the linux community - where your worth is measured
not by how much you accumulate, but by how much you give away. The
interesting thing is that guys who give away their tools tend to get
famous, and then, if the tools are any good, they make money on their
books. Larry Wall is a best-selling (for compsci books) author now, I
think, with "Programming Perl"; Terence Mann's book on PCCTS is also
out (I just got my copy from There are some people who
are making pretty good money teaching courses on Perl. Your tool will
never get to that stage, because not many of us will look at it, as
its competitors are free, and good.

I got Yacc++ because I needed an excellent tool for a commercial
client, and money was no object to them. I could never have afforded
it myself.

Another approach is to offer free source code for the complete system
to linux users, and only to linux users. Then sell the system to
others at what the market will bear. This marketing strategy is used
by the (excellent)QT GUI toolkit authors, who have the philosophy that
if you are using linux and you agree to make whatever you build
available to others for free, then you can use their GUI builder. If
you are going to make something commercial, then you have to pay for
their toolkit. It makes a kind of sense, if you think about it.

As a final comment, my current platform for building compilers is
Objective Caml. This is an amazingly powerful language whose
strengths seem to be just what you need for compiler writing, and
whose weaknesses don't seem to get in your way. It comes with rather
primitive lex and yacc facilities, but in combination with the
language itself, the toolkit provides cradle-to-grave support for
writing fast, efficient, easy-to-maintain language processors. The
more I work in this field, the less I like C/C++ as a tool. The
language level is wrong - you need garbage collection, first- class
functions, built-in lists and pattern matching, and so on, to write
compilers, and if you don't have them, you have to fake them with
kludges like STL, the Boehm collector, and so on. So any preprocessor
- including your tool for C/C++/Java - starts out at a disadvantage.
In my humble opinion.


Post a followup to this message

Return to the comp.compilers page.
Search the comp.compilers archives again.