I am trying to write a simple program using BDB , similar to the
programs given in the reference document. Both the key and data are
integers. For some reason, the DB->put function is failing (Invalid
Argument). I am not able to figure out what could fail with a simple
statement as that.
If I try to display the error using DB->err, I get a segmentation
fault.
And if I try to set the page (1K) and cache size (32K), those methods
fail as well.
Am I missing something important over here?
Thanks very much in advance.
Regards,
Maitreyee

rmaitreyee (AT) gmail (DOT) com (Maitreyee) wrote in message news:<db53bce6.0409131700.72d6c963 (AT) posting (DOT) google.com>...
Please post your code that will demonstrate the problem.

Regards,
Ron

Sleepycat Software

ron (AT) sleepycat (DOT) com (Ron Cohen) wrote in message news:<6906cc78.0409140628.c1695ec (AT) posting (DOT) google.com>...
Here is the piece of code:

#define SOMEDB "somedb.db"

int main(void)
{
DB *dbp;
DBT key,data;
int i;
int *pInt;
int *pData;
int ret;

if ( db_create( &dbp,NULL,0) != 0)
{
printf("Error creating database\n");
exit(0);
}
dbp->set_errfile(dbp, stderr);

if ((ret = dbp->set_pagesize(dbp, 1024)) != 0)
{
dbp->err(dbp, ret, "set_pagesize");
exit(0);
}
if ((ret = dbp->set_cachesize(dbp, 0, 32 * 1024, 0)) != 0)
{
dbp->err(dbp, ret, "set_cachesize");
exit(0);
}

if ( dbp->open ( dbp , NULL , SOMEDB , DB_RECNO ,DB_CREATE,
0664 ) != 0)
{
printf("Error opening client database\n");
exit(0);
}

if ( (pInt = (int*) malloc (sizeof(int))) == NULL )
{
printf("Error doing malloc for pInt\n");
exit(0);
}

for ( i=0 ; i<10 ; i++ )
{
memset ( &key , 0 , sizeof(key) );
memset ( &data , 0 , sizeof(data) );
*pInt = i+1;
key.data = pInt;
key.size = sizeof(int);
data.data = pInt;
data.size = sizeof(int);

if((ret = dbp->put(dbp,NULL,&key,&data,DB_APPEND)) !=
0 )
{
printf("Error putting record in the database for record :%d\n",i);
exit(0);
}
}
for (i=0;i<10;i+2)
{
*pInt = i;
key.data = pInt;
key.size = sizeof(int);

if ( dbp->get(dbp,NULL,&key,&data,0) != 0 )
{
printf("Error retrieving record\n");
exit(0);
}
pData = data.data;
printf("data is:%d",*pData);
}
free(pInt);
}

Running this as it is gives a segmentation fault which goes away with
the removal of dbp->err.
Thanks for your help.
Regards,
Maitreyee

rmaitreyee (AT) gmail (DOT) com (Maitreyee) wrote in message news:<db53bce6.0409141136.3b8c5734 (AT) posting (DOT) google.com>...
The DB->put function is due to a coding error. I modified your code
below.

Also, this program would not compile as it was missing an argument for
the get method.

I am not able to figure out what could fail with a simple

if ( dbp->open ( dbp , NULL , SOMEDB , NULL, DB_RECNO ,DB_CREATE,
0664 ) != 0)



I also changed the line above
for (i=1;i<10;i++)




May want to use:


key.flags = DB_DBT_USERMEM;



You are welcome.


Ron

Thanks Ron for your help.
Unfortunately, db->put is giving me trouble again.
I have a database and a secondary index associated with it.
Whenever I try to "put" to the primary, I get an "Unknown error". The
error number is a wierd (similar to -10056378200)negative value. If I
try to "put" to a database which does not have a secondary index
associated with it,it works super fine.Does any special arrangement
need to be done when the database is associated with a secondary?
Once again,Thanks in advance.
Regards,
Maitreyee

ron (AT) sleepycat (DOT) com (Ron Cohen) wrote in message news:<6906cc78.0409210635.6b20ac55 (AT) posting (DOT) google.com>...

Here is an example of using a secondary index in C. I hope this helps.

Ron
Sleepycat Software



#include <sys/types.h>

#include <assert.h>
#include <stdlib.h>
#include <string.h>

#include "db.h"

int getname(DB *, const DBT *, const DBT *, DBT *);
void insert(DB *, int, char *);

int
getname(dbp, pkey, pdata, skey)
DB *dbp;
const DBT *pkey, *pdata;
DBT *skey;
{
memset(skey, 0, sizeof(DBT));
skey->data = strchr(pdata->data, '.') + 1;
skey->size = strlen(skey->data);
return (0);
}

void
insert(dbp, id, name)
DB *dbp;
int id;
char *name;
{
DBT key, data;
char idbuf[10];

memset(&key, 0, sizeof(key));
memset(&data, 0, sizeof(data));

sprintf(idbuf, "%05d", id);
key.data = idbuf;
key.size = strlen(idbuf);

data.data = name;
data.size = strlen(name);

printf("insert: %s\n", name);
assert(dbp->put(dbp, NULL, &key, &data, 0) == 0);
}

int
main()
{
DB_ENV *dbenv;
DB *dbp, *sdbp;

(void)system("rm -rf TESTDIR; mkdir TESTDIR");

assert(db_env_create(&dbenv, 0) == 0);
dbenv->set_errfile(dbenv, stderr);
assert(dbenv->open(dbenv,
"TESTDIR", DB_CREATE | DB_INIT_MPOOL, 0666) == 0);

assert(db_create(&dbp, dbenv, 0) == 0);
assert(dbp->open(dbp, NULL,
"a.db", "primary", DB_BTREE, DB_CREATE, 0) == 0);

assert(db_create(&sdbp, dbenv, 0) == 0);
assert(sdbp->set_flags(sdbp, DB_DUP) == 0);
assert(sdbp->open(sdbp, NULL,
"a.db", "secondary", DB_BTREE, DB_CREATE, 0) == 0);

assert(dbp->associate(dbp, NULL, sdbp, getname, 0) == 0);

insert(dbp, 1, "Alice.AAAA");
insert(dbp, 2, "Bob.BBBB");
insert(dbp, 3, "Carl.CCCC");

assert(sdbp->close(sdbp, 0) == 0);
assert(dbp->close(dbp, 0) == 0);
assert(dbenv->close(dbenv, 0) == 0);

return (0);
}

rmaitreyee (AT) gmail (DOT) com (Maitreyee) writes:
Are you sure that the callback that you passed to DB->associate() is
*always* returning zero or a valid error number? DB itself doesn't use
such a large negative value for its errors, so it would seem that either
some C library call did so (doubtful) or your associate callback
returned that value.


Philip Guenther

Thank You both for your replies. I was not returning any value from
the callback routine.
I also realized that I was having a primary with duplicates , which as
the documentation says is not allowed when a secondary is to be
associated with it.
Would the following way be a potential workaround for the situation
where the primary has duplicates and yet a secondary is needed?
Step 1:Locate the first available key/data pair in the primary by
doing a pget on the secondary.
Step 2:Traverse the primary using a cursor to get the remaining
key/data pairs?

I am sorry if this is confusing.
Thanks again.
Regards,
Maitreyee


Philip Guenther <guenther (AT) lab (DOT) smi.sendmail.com> wrote