OID unsigned long long

* mlw <markw@mohawksoft.com> [010813 21:06]:

I am thinking about embarking on changing the typedef of OID to unsigned long
long.

My plan is to make it conditional at configure time, i.e.

#ifdef OID_ULONGLONG
typedef unsigned long long Oid;
#define OID_MAX ULLONG_MAX
#else
typedef unsigned int Oid;
#define OID_MAX UINT_MAX
#endif

Aside from adding %llu to all the %u everywhere an OID is used in a printf, and
any other warnings, are there any other things I should be specially concerned
about?

The wire protocol.......

LER

--
5-4-3-2-1 Thunderbirds are GO!
------------------------
http://www.mohawksoft.com

---------------------------(end of broadcast)---------------------------
TIP 4: Don't 'kill -9' the postmaster

--
Larry Rosenman http://www.lerctr.org/~ler
Phone: +1 972-414-9812 E-Mail: ler@lerctr.org
US Mail: 1905 Steamboat Springs Drive, Garland, TX 75044-6749

mlw <markw@mohawksoft.com> writes:

Aside from adding %llu to all the %u everywhere an OID is used in a
printf, and any other warnings, are there any other things I should be
specially concerned about?

FE/BE protocol, a/k/a client/server interoperability. Flagging a
database so that a backend with the wrong OID size won't try to run in
it. Alignment --- on machines where long long has to be 8-byte aligned,
TOAST references as presently constituted will crash, because varlena
datatypes in general are only 4-byte aligned. There are more, but that
will do for starters.

BTW, I think #ifdef would be a totally unworkable way to attack the
format-string problem. The code clutter of #ifdef'ing everyplace that
presently uses %u would be a nightmare; the impact on
internationalization files would be worse. And don't forget that %llu
would be the right thing on only some machines; others like %qu, and
DEC Alphas think %lu is just fine. The only workable answer I can see
is for the individual messages to use some special code, maybe "%O" for
Oid. The problem is then (a) translating this to the right
platform-dependent thing, and (b) persuading gcc to somehow type-check
the elog calls anyway.

regards, tom lane

Tom Lane wrote:

mlw <markw@mohawksoft.com> writes:

Aside from adding %llu to all the %u everywhere an OID is used in a
printf, and any other warnings, are there any other things I should be
specially concerned about?

FE/BE protocol, a/k/a client/server interoperability. Flagging a
database so that a backend with the wrong OID size won't try to run in
it. Alignment --- on machines where long long has to be 8-byte aligned,
TOAST references as presently constituted will crash, because varlena
datatypes in general are only 4-byte aligned. There are more, but that
will do for starters.

I will have to look at that, thanks.

BTW, I think #ifdef would be a totally unworkable way to attack the
format-string problem. The code clutter of #ifdef'ing everyplace that
presently uses %u would be a nightmare; the impact on
internationalization files would be worse. And don't forget that %llu
would be the right thing on only some machines; others like %qu, and
DEC Alphas think %lu is just fine.

What do you think of making two entries in the various printf strings, and
using macros to split up an OID, as:

printf("OID: %u:%u", HIGHOID(od) LOWOID(oid))

That may satisfy your concern for #ifdef's everywhere, and it could mean I
could submit my patches back without breaking any code, so PostgreSQL could be
closer to a 64 bit OID.

The only workable answer I can see
is for the individual messages to use some special code, maybe "%O" for
Oid. The problem is then (a) translating this to the right
platform-dependent thing, and (b) persuading gcc to somehow type-check
the elog calls anyway.

regards, tom lane

---------------------------(end of broadcast)---------------------------
TIP 1: subscribe and unsubscribe commands go to majordomo@postgresql.org

--
5-4-3-2-1 Thunderbirds are GO!
------------------------
http://www.mohawksoft.com

mlw writes:

I am thinking about embarking on changing the typedef of OID to unsigned long
long.

Aside from adding %llu to all the %u everywhere an OID is used in a printf, and
any other warnings, are there any other things I should be specially concerned
about?

You can start with my patch at

http://www.ca.postgresql.org/~petere/oid8.html

See the comments on that page and the other responses. It ain't pretty.

--
Peter Eisentraut peter_e@gmx.net http://funkturm.homeip.net/~peter

Tom Lane wrote:

[...]

BTW, I think #ifdef would be a totally unworkable way to attack the
format-string problem. The code clutter of #ifdef'ing everyplace that
presently uses %u would be a nightmare; the impact on
internationalization files would be worse. And don't forget that %llu
would be the right thing on only some machines; others like %qu, and
DEC Alphas think %lu is just fine. The only workable answer I can see
is for the individual messages to use some special code, maybe "%O" for
Oid. The problem is then (a) translating this to the right
platform-dependent thing, and (b) persuading gcc to somehow type-check
the elog calls anyway.

You can ask gcc to typecheck format strings for printf type functions
with something like the following:

extern int
my_printf (void *my_object, const char *my_format, ...)
__attribute__ ((format (printf, 2, 3)));

Ref: http://www.delorie.com/gnu/docs/gcc/gcc_77.html

David