PQgetvalue weirdness

Postgres numbers the rows and fields starting at 0 - your call to get the
first row and column would be PQgetvalue(result, 0, 0)

Chris

On Thu, 10 Sep 1998, Simon Drabble wrote:

On Thu, 10 Sep 1998, Chris Johnson wrote:

Postgres numbers the rows and fields starting at 0 - your call to get the
first row and column would be PQgetvalue(result, 0, 0)

Chris

Thanks for your reply, but..

If I have a zero for the field number (last parameter) I get a seg fault
immediately - gdb shows it to be in PQgetvalue().

If I have a row number of zero (second param) I get the same error as
below (if the column number is >0 )

Simon.

On Thu, 10 Sep 1998, Simon Drabble wrote:

I'm executing a (simple) query using libpq, and this is the error I get
from PQgetvalue(result, 1, 1):

PQgetvalue: There is no field 1 in the query results. The highest
numbered field is 0.

I get a similar error if I call getlength().

There _is_ data there - PQprintTuples() prints everything correctly.

Can anyone shed any light? PQtrace() doesn't reveal anything useful..

Simon.

--
...oooOOOooo...
Simon Drabble mailto:madlather@syspac.com
http://fiddletrudge.dyn.ml.org/
KreatAPayj -- ~simon/KreatAPayj/login.html
LizardCam -- LizardCam/LizardCam.html
Linux Knowledge Base -- ~simon/LinuxKB/index.html

--
...oooOOOooo...
Simon Drabble mailto:madlather@syspac.com
http://fiddletrudge.dyn.ml.org/
KreatAPayj -- ~simon/KreatAPayj/login.html
LizardCam -- LizardCam/LizardCam.html
Linux Knowledge Base -- ~simon/LinuxKB/index.html

I usually use pg_result... here is a snipped of some C I have used:

<<<>>>
for (i = 0; i < PQntuples(res); i++)
{
dbName = PQgetvalue(res, i, 0);
<<<>>>

It works perfectly on the Linux, IRIX and Digital Unix (OSF/1)... Assuming
all the regression tests worked when you installed, and that you are
linking your program with the proper libpq, I guess the next step is to
see more of the code and to know exactly how you compiled it.

Chris

--
A language that doesn't affect the way you think about programming is
not worth knowing.

On Thu, 10 Sep 1998, Simon Drabble wrote:

On Thu, 10 Sep 1998, Chris Johnson wrote:

I usually use pg_result... here is a snipped of some C I have used:

<<<>>>
for (i = 0; i < PQntuples(res); i++)
{
dbName = PQgetvalue(res, i, 0);
<<<>>>

It works perfectly on the Linux, IRIX and Digital Unix (OSF/1)... Assuming
all the regression tests worked when you installed, and that you are
linking your program with the proper libpq, I guess the next step is to
see more of the code and to know exactly how you compiled it.

I'm running Linux (RH4.2, 2.0.34) , latest version of pg (6.3.2?) and
linking AFAIK with
the correct libpq.a - I only have the one copy on my system so it should
be correct. The code is below. Everything else works but for the
getvalue()

dbresult is just a list-type structure with char* and dbresult* members.
dbconn is pre-initialised in another function.

Note: code has been stripped of non-essential stuff.

/* This function returns a _single_ column value from a table
* The return value is a list to allow multiple results
* When you've finished with the returned list, call db_freevalue(list)
*/
dbresult * db_getvalue(char * table, char * column, char * where_clause)
{
dbresult *values, *v;
char query[1024]; /* max is 8192 */
PGresult *result;
ExecStatusType status;
int tuples;
int count = 0;

if (!table || !column) return NULL;

sprintf(query, "SELECT %s FROM %s", column, table);

if (where_clause) sprintf(query, "%s %s", query, where_clause);

result = PQexec(dbconn, query);

if ((status = PQresultStatus(result)) != PGRES_TUPLES_OK) {

char temp[1024];
sprintf(temp, "SELECT returned %s\n Query was: %s",
PGerror[status], query);
/* logerr("DB", temp); */

return NULL;
}

tuples = PQntuples(result);

values = malloc(sizeof(dbresult));
v = values;

count = 0;

PQprintTuples(result, stderr, 1, 1, 0);
while (count <= tuples) {

/*
v->data = malloc(PQgetlength(result, count, 1));
*/
v->data = malloc(256);
v->next = NULL;

strcpy(v->data, PQgetvalue(result, count, 1));

if (--count > 0) {
v->next = malloc(sizeof(dbresult));
v = v->next;
}
}

PQclear(result);
return values;

}

Simon.

Chris

--
A language that doesn't affect the way you think about programming is
not worth knowing.

On Thu, 10 Sep 1998, Simon Drabble wrote:

On Thu, 10 Sep 1998, Chris Johnson wrote:

Postgres numbers the rows and fields starting at 0 - your call to get the
first row and column would be PQgetvalue(result, 0, 0)

Chris

Thanks for your reply, but..

If I have a zero for the field number (last parameter) I get a seg fault
immediately - gdb shows it to be in PQgetvalue().

If I have a row number of zero (second param) I get the same error as
below (if the column number is >0 )

Simon.

On Thu, 10 Sep 1998, Simon Drabble wrote:

I'm executing a (simple) query using libpq, and this is the error I get
from PQgetvalue(result, 1, 1):

PQgetvalue: There is no field 1 in the query results. The highest
numbered field is 0.

I get a similar error if I call getlength().

There _is_ data there - PQprintTuples() prints everything correctly.

Can anyone shed any light? PQtrace() doesn't reveal anything useful..

Simon.

--
...oooOOOooo...
Simon Drabble mailto:madlather@syspac.com
http://fiddletrudge.dyn.ml.org/
KreatAPayj -- ~simon/KreatAPayj/login.html
LizardCam -- LizardCam/LizardCam.html
Linux Knowledge Base -- ~simon/LinuxKB/index.html

--
...oooOOOooo...
Simon Drabble mailto:madlather@syspac.com
http://fiddletrudge.dyn.ml.org/
KreatAPayj -- ~simon/KreatAPayj/login.html
LizardCam -- LizardCam/LizardCam.html
Linux Knowledge Base -- ~simon/LinuxKB/index.html

--
...oooOOOooo...
Simon Drabble mailto:madlather@syspac.com
http://fiddletrudge.dyn.ml.org/
KreatAPayj -- ~simon/KreatAPayj/login.html
LizardCam -- LizardCam/LizardCam.html
Linux Knowledge Base -- ~simon/LinuxKB/index.html