[OpenSIPS-Devel] Db interface

[Stanisław Pitucha]

Hi,
I had another go at a new module and I'm confused about this
explanation. What does "passing db_res struct" mean in these contexts:

On 4 August 2010 21:17, Bogdan-Andrei Iancu <bogdan at voice-system.ro> wrote:
> 1) directly, by passing the db_res struct to the query function - it
...
> 2) piece by piece, after the query is done - you do not pass any db_res
> to the query function; after the query is done, you use fetch_result()

Does this mean, db_text is not compliant / not complete if it's code is:

--->8---
int dbt_query(db_con_t* _h, db_key_t* _k, db_op_t* _op, db_val_t* _v,
         db_key_t* _c, int _n, int _nc, db_key_t _o, db_res_t** _r)
{
[...]
   if ((!_h) || (!_r) || !CON_TABLE(_h))
   {
      LM_ERR("invalid parameters\n");
--->8---

If it fails when `(!_r)`, it seems like you cannot use the second method.
Or did you mean something else by "passing db_res" - allocating *_r,
but not _r itself?
My guess at what I should do is something like this for the query function:

--->8---
// do the query bits, get back the `cursor`
if (*_r) {
   db_allocate_columns(*_r, ...);
   db_allocate_rows(*_r, ...);
   .... // fill in the rows and columns
} else {
   SOME_CUSTOM_FIELD(_h) = cursor;
}
--->8---

Then again this doesn't make much sense to me. If that would be the
case, why is db_res_t** needed at all? If it's always allocated before
this function, why is it not db_res_t* instead?


Another separate issue I run into lately is, what should be filled in
the result in the second scenario. If I'm using a database which uses
only cursors and doesn't know the size of resultset before returning
it all, what do I do about the db_res->n and db_res->res_rows fields?
Can they be left not filled? And how does fetch_result signal that
there are no new results to fetch?

Regards,
Completely confused,
Stan