Re: contrib/intarray/_int_gist.c

In "g_int_compress" :

int *dr;
...
memmove((void *) &dr[cand - 1], (void *) &dr[cand + 1], (len - cand -
1) * sizeof(int));

Should be

int32 *dr;
...
memmove((void *) &dr[cand - 1], (void *) &dr[cand + 1], (len - cand -
1) * sizeof(int32));

<jw(dot)pgsql(at)sduept(dot)com> wrote
> In "g_int_compress" :
>
>
> int *dr;
> ...
> memmove((void *) &dr[cand - 1], (void *) &dr[cand + 1], (len - cand -
> 1) * sizeof(int));
>
> Should be
>
> int32 *dr;
> ...
> memmove((void *) &dr[cand - 1], (void *) &dr[cand + 1], (len - cand -
> 1) * sizeof(int32));
>

AFAICS, int32 and int are exactly the same thing in PostgreSQL. For the
machine int is not 32 bits long, PostgreSQL won't even run.

Regards,
Qingqing

"Qingqing Zhou" <zhouqq(at)cs(dot)toronto(dot)edu> writes:
> AFAICS, int32 and int are exactly the same thing in PostgreSQL. For the
> machine int is not 32 bits long, PostgreSQL won't even run.

Ideally we should operate correctly if "int" is 64 bits. In practice
I agree that making contrib work would be mighty far down the list of
things to fix...

It appears to me that the current de-facto standard for C on 64-bit
machines is
char 8 bits
short 16 bits
int 32 bits
long 64 bits
Promoting "int" to 64 bits has a big problem: you have to drop one of
the widths entirely, because there is no other basic type allowed by
C. (int16_t and the others are only typedefs not new basic types.)
So I'm not really expecting to see int = 64 bits any time soon.

As for the other direction (int = 16 bits), there's no real hope of
running Postgres on a 16-bit machine anyway :-(

regards, tom lane

Tom Lane wrote:
> "Qingqing Zhou" <zhouqq(at)cs(dot)toronto(dot)edu> writes:
> > AFAICS, int32 and int are exactly the same thing in PostgreSQL. For the
> > machine int is not 32 bits long, PostgreSQL won't even run.
>
> Ideally we should operate correctly if "int" is 64 bits. In practice
> I agree that making contrib work would be mighty far down the list of
> things to fix...
>
> It appears to me that the current de-facto standard for C on 64-bit
> machines is
> char 8 bits
> short 16 bits
> int 32 bits
> long 64 bits
> Promoting "int" to 64 bits has a big problem: you have to drop one of
> the widths entirely, because there is no other basic type allowed by
> C. (int16_t and the others are only typedefs not new basic types.)
> So I'm not really expecting to see int = 64 bits any time soon.
>
> As for the other direction (int = 16 bits), there's no real hope of
> running Postgres on a 16-bit machine anyway :-(

Agreed. CVS change made for clarity, int->int32.

--
Bruce Momjian http://candle.pha.pa.us
EnterpriseDB http://www.enterprisedb.com

+ If your life is a hard drive, Christ can be your backup. +