General talk about software patterns

[Ralph Johnson <johnson AT cs.uiuc.edu>]

On 10/24/04 6:10 PM, "Mark Grand" 
<mgrand AT mindspring.com>
 wrote:

> In the commercial sector, it is hard to find Lisp or Smalltalk people.
> Reimplementing in a more common language will, over time, reduce costs and
> make the progress of software development more predictable.

This is what managers think.  Managers are wrong.  Converting a Lisp or
Smalltalk program always increases costs and makes the progress of software
development dramatically less predictable.  It isn't even close.

I've seen Smalltalk projects in trouble because the programmers all left.
This was mostly back in the days when Smalltalk was very hot and there were
programmers making $2000 a day.  (Not many, but a few.)  It was hard to keep
good programmers on a project, especially with bad management.  Managers
didn't have that problem with COBOL or C.  But now Smalltalk is not in such
demand.  Salaries are the same as for other languages and Smalltalk
programmers don't move around any more than other programmers.

It is hard to hire a lot of Smalltalkers fast, but it is easy to hire them
slow.  So, you shouldn't try to have a 100 person Smalltalk project, but
those were always bad ideas.  Part of the magic of Smalltalk is that you can
do with ten people what would take 100 with another language.

It is well known that the key to successful software projects is getting
good people.  Good people prefer productive programming languages.  In my
experience, the best programmers prefer "weird" languages like Smalltalk or
Lisp.  Or Ruby, or Ocaml.  Productive programming languages let your
development group be smaller, reducing management problems.

-Ralph Johnson