[erlang-questions] "Design patterns" for functional languages?

[Michael T. Richter]

On Wed, 2008-08-06 at 19:34 -0700, Kevin Scaldeferri wrote:

> > These would fall more under the heading of "idioms" than GoF-style  
> > patterns in my books (although the latter of the two is in the same  
> > grey space I mention below).



> Indeed, it's tough to say exactly what the distinctions are between  
> "idioms", "design patterns", and "best practices".  And what is idiom  
> in one language may be a pattern in another.


Well, I guess it's the same as the difference between tactical,
operational or strategic command: scope of focus.

To me an "idiom" involves a pretty local, narrow focus.  The canonical C
++ example is the auto_ptr while the canonical Ruby example would be the
||= initialization trick.  A "pattern" on the other hand has broader
scope that impacts the way the whole is built and functions.  Even the
Singleton pattern that I squint at as something that is perhaps not
quite a "pattern" in my books impacts several things application-wide
(specifically the way one or more objects are made).  Another one that I
squint at suspiciously -- the Class Factory -- falls too into this.  Use
of this pattern affects the way everything else that interacts with it
has to be coded.  This is in no way the case (except for the missing
bugs, naturally) when you look at auto_ptr or ||=.

As to "best practices" -- I find those to be generally social constructs
that impact code, not architectural elements.

But yes, I agree fully that an idiom in one language is a pattern in
another.  The GoF Iterator and Visitor are such beasts.  In C++ or Java
they're patterns.  In Eiffel or Dylan they're idioms.  In Haskell
they're so trivial you don't think about them as anything special beyond
"it's the way you do it".

The OP did bring up an interesting point: I have to confess that I now
disagree with the reference I sent that said you don't need GoF-style
patterns in the functional world.  The OTP is a collection of GoF-style
patterns for Erlang.  It's just way more abstract and sophisticated a
pattern and I failed to see the forest for the trees as a result.
Monads in Haskell are the same deal, as are the various
continuation-passing approaches in various functional languages.  I
stand, hereby, corrected.

-- 
Michael T. Richter <ttmrichter@REDACTED> (GoogleTalk:
ttmrichter@REDACTED)
Never, ever, ever let systems-level engineers do human interaction
design unless they have displayed a proven secondary talent in that
area. Their opinion of what represents good human-computer interaction
tends to be a bit off-track. (Bruce Tognazzini)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://erlang.org/pipermail/erlang-questions/attachments/20080807/f33ad482/attachment.htm>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 189 bytes
Desc: This is a digitally signed message part
URL: <http://erlang.org/pipermail/erlang-questions/attachments/20080807/f33ad482/attachment.bin>