General talk about software patterns

[Malte Finsterwalder <malte AT finsterwalder.name>]

Ralph Johnson wrote:

> This is a typical story.  Why do people take successful systems in
> Lisp (or Smalltalk, or ...) and rewrite them in much more boring and less
> powerful languages?  We need to understand this if we want to make the world
> safe for powerful languages.

I think analyzing the history of Java can give us a lot of information
about how to make a language successfull.

Analyzing the "failures" of other languages can give us even more
information.

I think the success of Java is due to:
1) Good timing - the Internet hype just started and Java was for the
Internet
2) Good marketing - Sun made Java heard and told everyone how great it is
3) Free tools - the jdk was free right from the start and now even
complete development environments are free (eclipse, netbeans...)
4) Standardized but not by committee - There is only one Java and
changing the language standard and the library happens more quickly than
in committee-based languages, like C++ for example
5) There is a very large standardized library for most of the general
purpose programming tasks. It may not always be perfect or even very
good, but it's there
6) Platform independent - I think that by now this is the most
interesting feature of Java
7) It head a smooth migration path for a lot of programmers because of
it's similarities to C/C++

Why did other languages fail?
1) Not so much marketing and no support by very big companies
2) No Standard and a lot of different implementations (e.g. Smalltalk)
3) Rather expensive tools make it hard to even try the language
4) A standard that is developed too slow and the implementations don't
even fully support it (C++)
5) Even for the most general applications you need additional libraries
to get something done (C++). GUIs are a particular problem here, I
think. Most Languages don't support GUIs without external tools and
libraries.
6) They look freaky or complicated and are hard to read and/or write for
beginners (e.g. Lisp)

Looking at this I see some patterns:

It's a mixture of technical features and marketing, that makes a
language successfull.
The design of the language itself is not important. You can't screw the
design up completely of course, but you don't need to get it
exceptionally well either. Just a little better than what most people
are used to.

Underlying all this is: To market a language you need a good business
case. You need a problem that your language solves better than all the
other languages. At least you need to make people believe that.

What do you think?

Greetings,
   Malte