Copying one Lisp_Object to another in C code?

Copying one Lisp_Object to another in C code?

[James Cloos]

Given:

    static void (foo)
        Lisp_Object foo;
    {
        Lisp_Object bar;
    /* etc */

is it OK to do:

        bar = foo;

or is there a function or macro one should call?

I presume GCPRO1 (foo) is in order either way?

-JimC
-- 
James Cloos <cloos@jhcloos.com>         OpenPGP: 1024D/ED7DAEA6

Re: Copying one Lisp_Object to another in C code?

[Ken Raeburn]

On Jul 1, 2010, at 20:18, James Cloos wrote:
> Given:
> 
>    static void (foo)
>        Lisp_Object foo;
>    {
>        Lisp_Object bar;
>    /* etc */
> 
> is it OK to do:
> 
>        bar = foo;
> 
> or is there a function or macro one should call?

Simple assignments are done all over the place currently.  It would take some drastic reworking if some macro invocation were to become necessary.

> I presume GCPRO1 (foo) is in order either way?

Depending on the details of your usage, yes.  (If the value is always stored in other places where the version of the GC system that doesn't scan the entire stack and register set automatically will be able to find it, at any of the points where the GC system can be invoked, then you don't need an explicit GCPRO.  But then you have to understand where the old GC scanner will look, and when it can get invoked.)  Just being conservative about it shouldn't hurt, either.

Ken

Re: Copying one Lisp_Object to another in C code?

[Chong Yidong]

James Cloos <cloos@jhcloos.com> writes:

> Given:
>
>     static void (foo)
>         Lisp_Object foo;
>     {
>         Lisp_Object bar;
>     /* etc */
>
> is it OK to do:
>
>         bar = foo;

Yes.

> I presume GCPRO1 (foo) is in order either way?

Only if the code ends up calling eval, and for the benefit of the
platforms that don't use stack marking.

Re: Copying one Lisp_Object to another in C code?

[Andreas Schwab]

James Cloos <cloos@jhcloos.com> writes:

> Given:
>
>     static void (foo)
>         Lisp_Object foo;
>     {
>         Lisp_Object bar;
>     /* etc */
>
> is it OK to do:
>
>         bar = foo;
>
> or is there a function or macro one should call?

Lisp_Object values are just references (except for Lisp_Int values),
so they can be freely copied around.

> I presume GCPRO1 (foo) is in order either way?

Only if you reference a Lisp object that is not already protected from
elsewhere.  For example, all argument values of built-ins are implicitly
protected.

Andreas.

-- 
Andreas Schwab, schwab@linux-m68k.org
GPG Key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."

Re: Copying one Lisp_Object to another in C code?

[James Cloos]

Thanks all!

-JimC
-- 
James Cloos <cloos@jhcloos.com>         OpenPGP: 1024D/ED7DAEA6

Re: Copying one Lisp_Object to another in C code?

[Stefan Monnier]

> I presume GCPRO1 (foo) is in order either way?

You may either need GCPRO1(foo) or GCPRO1(bar) or GCPRO2(foo, bar), or
nothing at all.  It all depends on what the rest of the code does.


        Stefan