Sequence point question

[Gene]

Hello all. I do not have access to the ANSI standard doc. Perhaps
someone can help me with this. Suppose I have this situation:

typedef void *PTR;

struct cons_s {
PTR car, cdr;
};

struct root_record_s {
...
struct cons_s *p;
};

PTR cons(struct root_record_s *rr, PTR car, PTR cdr)
{
...
rr->p->cdr = gc_malloc(rr, sizeof(struct cons_s));
...
}

where gc_malloc may change the value of rr->p. Is there anything in
the ANSI standard that says rr->p->cdr will be the "correct" l-value,
e.g. the new one after gc_malloc() makes the change? Or must the
assignmet be split into two to ensure correct sequencing? ...

PTR cons(struct root_record_s *rr, PTR car, PTR cdr)
{
PTR tmp;
...
tmp = gc_malloc(rr, sizeof(struct cons_s));
rr->p->cdr = tmp;
...
}

Many thanks.

[Thad Smith]

Gene wrote:

rr->p->cdr = gc_malloc(rr, sizeof(struct cons_s));

where gc_malloc may change the value of rr->p. Is there anything in
the ANSI standard that says rr->p->cdr will be the "correct" l-value,
e.g. the new one after gc_malloc() makes the change?

No.

Or must the
assignmet be split into two to ensure correct sequencing?

Yes. Even though calling a function results in sequences points, the
point at which rr->p is evaluated is unspecified with regard to when the
function is called.

--
Thad

[Richard Heathfield]

Gene said:

Hello all. I do not have access to the ANSI standard doc. Perhaps
someone can help me with this. Suppose I have this situation:

<snip>

PTR cons(struct root_record_s *rr, PTR car, PTR cdr)
{
...
rr->p->cdr = gc_malloc(rr, sizeof(struct cons_s));
...
}

where gc_malloc may change the value of rr->p. Is there anything in
the ANSI standard that says rr->p->cdr will be the "correct" l-value,
e.g. the new one after gc_malloc() makes the change?

No, 'fraid not. As you thought, you have to use a temp to be sure of the
code doing what you want.

--
Richard Heathfield <http://www.cpax.org.uk>
Email: -www. +rjh@
Google users: <http://www.cpax.org.uk/prg/writings/googly.php>
"Usenet is a strange place" - dmr 29 July 1999

[Flash Gordon]

Gene wrote, On 23/06/07 05:20:

Hello all. I do not have access to the ANSI standard doc. Perhaps

<snip>

If you go to http://clc-wiki.net/wiki/c_standard there are links to
where you can obtain drafts of the various versions of the C standard
and information about the different types, including why I use the ISO
(or BSI) standard rather than the ANSI one.

Others have answered your main question correctly so I won't.
--
Flash Gordon

[Roland Pibinger]

see also:
http://www.embedded.com/shared/print...icleID=9900661
--
Roland Pibinger
"The best software is simple, elegant, and full of drama" - Grady Booch

[Gene]

On Jun 23, 5:18 am, Flash Gordon <[e-mail]:

Gene wrote, On 23/06/07 05:20:

Hello all. I do not have access to the ANSI standard doc. Perhaps

<snip>

If you go tohttp://clc-wiki.net/wiki/c_standardthere are links to
where you can obtain drafts of the various versions of the C standard
and information about the different types, including why I use the ISO
(or BSI) standard rather than the ANSI one.

Others have answered your main question correctly so I won't.
--
Flash Gordon

Thanks! This is very helpful, as are all the answers to the original
question.