[erlang-questions] My biggest beef with Erlang
Mon Nov 26 09:43:15 CET 2007
Kenneth Lundin wrote:
> I don't think improvements in support for parse transforms and a broader use of
> parse transforms is one of the most important bits in this.
> I even think that it might be negative to broaden the use of parse transforms.
> Parse transforms is a way to let developers modify the language or even
> define their own new language, when you see a program you will have a hard time
> understanding what it means etc. It will be problems for syntax
> sensitive editors
I think that that is a crucial point. A good language should have a very small
number of primitives and rules. Everything you add to it is a potential increase
of complexity. It is true that some times a new primitive can reduce the cost of
coding---and understanding the code---, so the addition pays off the extra
complexity. But if you keep adding things that are not really needed you may end
in something like perl: you can write your code almost in any form you like, but
hardly read code written by others ...
For me, user defined operators and bang-bang rpc operator are things to be
discussed (I am not particularly fond of them), but there are other parts of the
language, library, and tools that need much more critical improvements.
For example, type checking adds more syntax, but to do something that cannot be
done right now, bitstreams add new functionalities with almost no changes of the
language, we need a _right_ solution for records, etc.
More information about the erlang-questions