BUG #11761: range_in dosn't work via direct functional call

On 10/22/2014 08:27 PM, olegjobs@mail.ru wrote:

This function test_ext_get_range(cstring) returns int4range:

Datum test_ext_get_range(PG_FUNCTION_ARGS)
{
char *ts = PG_GETARG_CSTRING(0);

return DirectFunctionCall3(range_in, CStringGetDatum(ts),
ObjectIdGetDatum(3904), Int32GetDatum(0);
}
In psql:
select test_ext_get_range('[1,1]');
error:
connection to the server was lost.

it seems to me that some memory problems because of "The range I/O functions
need a bit more cached info than other range
* functions, so they store a RangeIOData struct in fn_extra, not just a
* pointer to a type cache entry. "

Yeah, DirectFunctionCall cannot be used with range_in. Use FunctionCall3
instead. See this comment in fmgr.c, above DirectFunctionCall1Coll:

/*
* These are for invocation of a specifically named function with a
* directly-computed parameter list. Note that neither arguments nor result
* are allowed to be NULL. Also, the function cannot be one that needs to
* look at FmgrInfo, since there won't be any.
*/

range_in needs to look at FmgrInfo.

- Heikki

--
Sent via pgsql-bugs mailing list (pgsql-bugs@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-bugs

2014-10-24 15:08 GMT+02:00 Heikki Linnakangas <hlinnakangas@vmware.com>:

On 10/22/2014 08:27 PM, olegjobs@mail.ru wrote:

This function test_ext_get_range(cstring) returns int4range:

Datum test_ext_get_range(PG_FUNCTION_ARGS)
{
char *ts = PG_GETARG_CSTRING(0);

return DirectFunctionCall3(range_in, CStringGetDatum(ts),
ObjectIdGetDatum(3904), Int32GetDatum(0);
}
In psql:
select test_ext_get_range('[1,1]');
error:
connection to the server was lost.

it seems to me that some memory problems because of "The range I/O
functions
need a bit more cached info than other range
* functions, so they store a RangeIOData struct in fn_extra, not just a
* pointer to a type cache entry. "

Yeah, DirectFunctionCall cannot be used with range_in. Use FunctionCall3
instead. See this comment in fmgr.c, above DirectFunctionCall1Coll:

There is a special "InputFunctionCall"

Regards

Pavel

Thanks for help, for other people I leave some snippet,  it took one working day to dig into postgres code to understand this:

FmgrInfo fmgr;

Oid func_id = fmgr_internal_function("range_in");
fmgr_info(func_id, &fmgr);

Datum result = FunctionCall3(&fmgr, CStringGetDatum("(1,1)"), ObjectIdGetDatum(3904), Int32GetDatum(0));
And one more question, for example, I do some memory allocation via palloc in _PG_Init().
Is this memory preserve for all life time after library is loaded into postgresql process?

Other case, for example, I have this code in dynamic lib:

char *global_init;
char *global;

void _PG_init(void)
{
   global_init = (char *) palloc(10);
}

Datum f (PG_FUNCTION_ARGS)
{
   if (!global)
   {
      global  = (char*) palloc (10);
   }
   *global = '\0'; // is it ok when i call "f" several times?
   *global_init = '\0'; // is it ok when i call "f" several times?
} 

Fri, 24 Oct 2014 15:52:36 +0200 от Pavel Stehule <pavel.stehule@gmail.com>:

=?UTF-8?B?RHVuYXVza2FzIE9sZWc=?= <olegjobs@mail.ru> writes:

And one more question, for example, I do some memory allocation via palloc in _PG_Init().
Is this memory preserve for all life time after library is loaded into postgresql process?

Probably not --- CurrentMemoryContext would most likely be pointing at a
statement-lifespan context. If you want that behavior, I'd recommend
explicitly allocating in TopMemoryContext.

(Also, you would not be needing to ask that question if you were using
an --enable-cassert build for testing. Which is *highly* recommended
when developing C code for Postgres.)

regards, tom lane

--
Sent via pgsql-bugs mailing list (pgsql-bugs@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-bugs