[fdo-dev] About DBI, GDBI and RDBI

[Mateusz Loskot]

Mateusz Loskot wrote:
> 
> Next, I'd like to ask for some suggestion about how to implement
> the cursor structure for PostGIS.
> I analysed MySQL and ODBC versions but both are completely different
> and not comparable, so may be you could provide me with general rules
> about Rdbi cursor implementation best practice?
> 
> Here is my prototype:
> 
> typedef struct tag_postgis_redefine_def
> {
>     void *original;
>     void *substitution;
>     void *geometry; /* Geometry to be deleted, since the client won't */
>     int   orig_type;
> } postgis_define_def;
> 
> As you see, my postgis_define_def is defined as MySQL's version.
> Now, the cursor:
> 
> typedef struct tag_postgis_cursor_def
> {
>     PGresult *statement;
>     int   define_count;
>     PGSQL_BIND *defines;
>     int   bind_count;
>     PGSQL_BIND *binds
>     postgis_define_def *redefines; /* array of size define_count, NULL
> entries for non-geometries */
>     postgis_define_def *rebinds;   /* array of size bind_count, NULL
> entries for non-geometries */
> } postgis_cursor_def;
> 
> PGresult* statement is pointer to prepared statement
> returned by PQprepare() function, from PostgreSQL API.
> So, this statement includes name, SQL query.


Please, forget this question and structures above.
Yesterday, I found much better solution, hopefully after
I've understood some missing rdbi details :-)

Cheers
-- 
Mateusz Loskot
http://mateusz.loskot.net