On Tue, Sep 8, 2009 at 9:59 AM, ?iga Lenar?i?<ziga.lenarcic at gmail.com> wrote:
> On Sep 8, 2009, at 3:22 PM, Stavros Macrakis wrote:
>> It is an error to use Lisp's push function to modify a Maxima
>> variable. In particular, this means you can't push onto a subscripted
>> variable: zz[1]:[]$ push(3,zz[1]) => error.
>
> Ah, these are the cases I always forget about, because accessing a maxima
> variable that is also a lisp variable is so easy to do. So in this case I
> can get the attached list with 'meval', modify it, but how do I then attach
> it back to the zz[1]?
Good question. Another interesting question is whether the push
variable should be evaluated twice. Consider:
push( 3, x[i:i+1] )
Surely you don't want this to be equivalent to
x[i:i+1] : cons(3, x[i:i+1])
which assigns to a different array location than the one from which it
reads, but rather something like
( temp: (i:i+1), x[temp]: cons(3, x[temp] ) )
This sort of mess is perhaps one reason that push is not part of the
standard Maxima core.
> One way is probably by constructing a 'mset' expression and evaluating it,
> but can we use something cheaper in this case (since it is known what we
> want)?
Why would you construct an mset expression? There is already an mset
function in mlisp.lisp.
> To make a fast push/pop I would still have special code, when push/pop gets
> a 'symbolp' (regular Maxima variable). In this case it's probably OK to just
> use lisp's 'set'?
This is a bad idea for two reasons: 1) it makes your code unnecessary
complicated; 2) it duplicates the same functionality in two places.
The second is the really crucial point. If at some point the
semantics of mset change (e.g. because we change scoping rules or
whatever), how will the maintainer know that he/she should also change
your push/pop functions? And are they really that slow?
-s