[EMAIL PROTECTED] writes:
>
> Christophe Rhodes writes:
> > On Mon, Dec 09, 2002 at 12:35:17PM -0600, [EMAIL PROTECTED] wrote:
> > > I've been getting some cryptic error messages from the following block
> > > of code with CMUCL 18d
> > >
> > > (in-package "COMMON-LISP-USER")
> > >
> > > (defmacro setassoc (key new-element alist &key (test #'eq))
> > ^^^^^^^^^^^
> >
> > Here's your problem -- using this you'll try to dump the literal
> > function object to file, rather than a reference to a function. You
> > want (test '#'eq) there instead.
>
> Thanks. I've made the change and that seems fine. But I was
> wondering if anyone could clarify for me why this happens in macros
> and does not happen when I use #'eq in function definitions?
>
> The same file has several functions that have #'eq as an argument, but
> they work fine.
Because the definition of those functions contain the source form
(function eq), not the literal function object #<Function EQ {10007891}>.
The macro looks up the EQ function at compile-time, then tries to use
that literal object in the source, which you can't dump in the FASL
file. If the macro expands into a run-time lookup of the EQ function,
it's all gravy.
--
/|_ .-----------------------.
,' .\ / | No to Imperialist war |
,--' _,' | Wage class war! |
/ / `-----------------------'
( -. |
| ) |
(`-. '--.)
`. )----'
