[antlr-interest] ST 4.0 planning
mail at martin-probst.com
Thu Sep 11 01:53:42 PDT 2008
>> Do I give a rat's ass if my text is generated in a functional
>> language? At the time, I
>> was trying to implement automatically generated sequence numbers
>> for things
>> like test cases and symbols. It was far more difficult than it
>> needed to be
> I especially like your observation about languages. The design of the
> language should not get in the way of getting common stuff done
> (Prolog, F#,
> ERLANG, etc.), without giving even more back for the effort (as do C+
> Java, C#, etc.).
The point of languages like Erlang or Prolog is that they make
different design tradeoffs than languages like Java or C#, and this
leads to very compelling features, like the robustness and
restartability (eh, not a word...) of Erlang micro-processes. Building
in some "convenience features", like mutable variables, would
completely taint the language.
About functional languages: programming in a functional style is
something you need to get used to, and it takes time. However many
people find that programs written in a pure functional language have
significantly less bugs, among other advantages. And in particular
transformation problems are very amenable to such stuff.
Something like a global, mutable counter variable might be a nice
quick hack to get something done, but it fails on a lot of other
levels, e.g. modularity, ability to test, potentially thread safety,
and so on.
I'd really recommend you to give this stuff some more time, it is
actually useful and a good idea.
More information about the antlr-interest