Re: setf.scm

[Mikael Djurfeldt <mdj@nada.kth.se>]

Christian Lynbech <chl@tbit.dk> writes:

> >>>>> "Mikael" == Mikael Djurfeldt <mdj@nada.kth.se> writes:
> 
> Mikael> Now I'm thinking of letting #:accessor <name> create the accessors
> Mikael> <name> and set-<name>! instead.
> 
> I am not happy with set-<name>!. One has to be carefull about
> polluting peoples namespace. So either accessors should be kept as
> generic functions, extending the getter name to also handle setting,

What do you mean?

  (x o 1)
  (x o) -> 1

?

I think

  (set-x! o 1)
  (x o) -> 1

is clearer.

> What I like about the setf! approach is that it is extremely usefull
> against arbitrary namespace pollution. For instance, records could
> really benefit from this as well.

I don't think it is arbitrary.  And, besides, if the user isn't happy
about set-x! he can choose a better name himself with #:setter, and he
doesn't even need to create getter or setter names at all.

We can export an STk-version of setf! (named `set!') from a
compatibility module together with a version of `define-class' which
uses an STk metaclass.

It may also be more consistent to rename #:accessor to #:accessors if
it's going to generate both `x' and `set-x!'.  (Logic: Both mutators
and selectors are "accessors".)

/mdj