Re: [Caml-list] OCaml's variables

[oliver <oliver@first.in-berlin.de>]

On Sat, May 25, 2013 at 12:30:12PM +0000, David Allsopp wrote:
> oliver wrote
> > Hi,
> > 
> > On Sat, May 25, 2013 at 01:49:46PM +0200, Gabriel Scherer wrote:
> > > The name "variable" has been used for centuries by the mathematicians
> > > to denote exactly names that refer to values (of course immutable
> > > values, we're talking about mathematical objects her). There is no
> > > point in trying to change it because some people conflate this notion
> > > with mutability.
> > 
> > Hmhhh.
> > "Some" people? As far as I remember, I heard these arguments quite often.
> 
> Some does not mean few. You appear to be trying to split hairs without a
> sufficient command of English. Once upon a time I would have prefixed that
> "With respect", but various Internet memes mean that most people now know what
> a British person actually means when they say that ;o)

The hairsplitting is built into myself - immutable property.
Regarding my english language skills I have to admit, that they could be better.


> 
> > > (When you write (x + 1), the value denoted by "x" is an unknown that
> > > depends on the context/environment; in particular it may vary when
> > > used in different contexts.)
> > >
> > > The name "variable" is also used in the expression "type variable",
> > > which denotes for example the 'a in
> > >   val id : 'a -> 'a
> > > and which is clearly not associated to any form of mutation.
> > [...]
> > 
> > Hmhhh, and what about imperative languages?
> > What part did they broke?
> > And why did they "misuse" the word "variable" and in which way?
> >
> > Is it because of mutability, which may break all these used "mathematical
> > metaphors"?
> > 
> > Can you elaborate on why "variables" are the wrong term or a 2bad" term in
> > imperative languages and not in functional languages?
> 
> There is no misuse and nothing broken - it is merely that when talking about
> imperative languages the word "variable" is typically where "mutable variable"
> is meant (because immutability is not usually of interest). When first teaching
> functional languages to people who are probably already familiar with
> imperative languages, one often chooses to eliminate the use of the word
> variable. This is because it can be confusing when trying to convey how to
> think functionally versus imperatively (a child's trick: if the student uses
> the word variable they are probably thinking of a reference cell and hence not
> functionally - so you steer the argument by "banning" the word variable from
> the conversation). Once you actually have an understanding of functional
> thinking, the elimination of the word variable becomes unnecessary.

Aha, I see.
Someone also brought in the argument of type variable.
I never had a problem with this so far.
I wonder, why I had a problem now with the stuff on page 136.
Maybe because type variable means, it's not fixed if it's type 'a,
but that a value (e.g. the recursive lists example on that page)
is a non-referenced and therefore immutable value.

The term for the input of a function normally is an "argument" or "parameter",
but seldom called variable.
Maybe thats why I stumbled over this term now.

> 
> The OCaml Language Specification is most certainly not a document intended
> for a beginner either to OCaml or functional programming - an obsessive
> elimination of the word variable would likely make the text less clear to those
> who will actually be reading it.

OK, I see.


Ciao,
   Oliver