Re: How hard is it to write a compiler ?

David Chase <chase@naturalbridge.com>
23 Oct 2000 22:11:28 -0400

          From comp.compilers

Related articles
How hard is it to write a compiler ? pradeeptumati@excite.com (Pradeep Tumati) (2000-10-22)
Re: How hard is it to write a compiler ? rbenedik@fsmat.htu.tuwien.ac.at (Ronald Benedik) (2000-10-23)
Re: How hard is it to write a compiler ? chase@naturalbridge.com (David Chase) (2000-10-23)
Re: How hard is it to write a compiler ? lionelp@worldonline.co.za (Lionel Pinkhard) (2000-10-23)
Re: How hard is it to write a compiler ? vbdis@aol.com (2000-10-23)
Re: How hard is it to write a compiler ? dlindauer@notifier-is.net (david lindauer) (2000-10-26)
Re: How hard is it to write a compiler ? jason@pattosoft.com.au (Jason Patterson) (2000-10-26)
| List of all articles for this month |

From: David Chase <chase@naturalbridge.com>
Newsgroups: comp.compilers
Date: 23 Oct 2000 22:11:28 -0400
Organization: Compilers Central
Keywords: practice

> Acutally, I am preparing a small business proposal for my management
> class. I will be happy to get your advices on this.


[It depends a great deal on the programmers. If you have really good
programmers, two of them should be able to do either project in a year.
If you don't, it'll take a dozen of them five years. -John]


I am not sure about your estimate; I think it depends upon what
you want the compiler to do, and the scope of the compiler. For
instance, writing a Modula-3 compiler and runtime and some of
the libraries took 3 full-time and 2 part-time people a little
over two years, that though included a couple of ports, plus
a mid-course correction when the language changed. But at least
two of those three full-time people (i.e., the two that weren't
me) were very good programmers.


On the other hand, simply rewriting the (optimizing) back-end of
an existing compiler, with the constraint of generating better
code than the back-end being replaced, took about five people
about three years. But, that was an optimizing back-end with
a target to beat; that adds more code, and more code means more
bugs, plus the target moved.


As a business proposal, I am not sure a compiler is a good idea,
unless you can find some way to not call it a compiler. Again, it
depends upon your goals. Venture capitalists are not terribly
interested in "tools", at least not last time I checked. One notable
exception is Curl who is pitching their product more along the lines
of "web content" than "tool", even though (to my amusement) it looks
rather much like LaTeX meets Lisp, but with marginally better syntax
all round (talk about faint praise :-). (As a Lisp and LaTeX fan, I
find this somewhat interesting, but if I'm a typical customer then
there's going to be some very disappointed investors.) There are a
couple of small compiler shops that never grew very large, but made a
decent living being small, at least as I understand it (EDG and
Silicon Valley Compilers come to mind).


If you have strict time-to-market concerns, you would be better
off buying parts (Edison Design Group sells a C/C++ front end,
I know that Declarative Systems once offered an attribute grammar
based system for C++, there is always GCC if the licensing doesn't
cause you trouble).


David Chase
chase@naturalbridge.com


Post a followup to this message

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