On Tuesday 22 March 2005 05:32 pm, Robert Dodier wrote:
> --- Barton Willis wrote:
> > Both tellsimp and tellsimpafter work by pattern matching;
> > the proposed post evaluation function doesn't. So the name
> > 'tellsimplast' might mislead users.
>
> I guess the concern I am trying to express is that
> we want to steer away from a new feature which
> seems out of harmony with the rest of Maxima. I confess
> I don't know what that implies about the name.
This isn't really a user "feature" - it's more of a programmer
feature.
> > Should the list of post evaluation functions be
> > a maxima list or a lisp list? If a user set this
> > list to a non-list, it could result in strange
> > error messages.
>
> Well, a general convention in Maxima seems to be that
> objects (functions, macros, rules, etc) are kept on a
> list (one of the "infolists"), but the list is maintained
> by one or more functions, and although the list can be
> inspected, it is not manipulated directly. I guess I'm
> inclined to suggest the same approach here. There is
> probably precedent for handling it some other way.
Hard to say - I suspect many precedents for many things could be
found by sufficient digging in the Maxima codebase ;-). I think,
given the intended uses of this feature, manipulating a list ala
Barton's latest proposal is the right way to go, particularly given
it can be protected using a standard Maxima mechanism I wasn't
aware of before. A programmer won't have a problem working with
the list directly, but it should make a regular user hesitate,
which is (IMHO) the desired result.
CY