Separate user and forum db's

Would that need to be done throughout the code itself, or should using that within the dblayer query function work?

i.e:

        function query($sql, $unbuffered = false)
        {
                $sql = str_replace($this->prefix.'users', 'userdb.'.$this->prefix.'users', $sql);
                $sql = str_replace($this->prefix.'groups', 'userdb.'.$this->prefix.'groups', $sql);
                $sql = str_replace($this->prefix.'online', 'userdb.'.$this->prefix.'online', $sql);

I've just tried that method within the query function, and it error logs as:

STATEMENT:  SELECT u.*, g.*, o.logged FROM userdb.users AS u INNER JOIN userdb.groups AS g ON u.group_id=g.g_id LEFT JOIN userdb.online AS o ON o.ident='[I.P address]' WHERE u.id=1
ERROR:  3F000: schema "userdb" does not exist
LOCATION:  LookupExplicitNamespace, namespace.c:1289
STATEMENT:  SELECT user_id FROM userdb.online WHERE idle=0
WARNING:  25P01: there is no transaction in progress
LOCATION:  EndTransactionBlock, xact.c:2931

Apologies if I'm missing the point again.

Thanks again,

Matt

He's probably using pgSQL.

It is indeed PgSQL.

With it not being a native function, (but I've found there is an addon to achieve the functionality, in a fashion), is there any plausible way of integrating this into the query:

http://www.postgresql.org/docs/8.3/stat … blink.html
http://www.postgresql.org/docs/8.3/static/dblink.html

By the looks of that documentation, str_replace wouldn't be upto the job for that syntax.

Thanks again,

Matt

$sql = str_replace($this->prefix."users", "dblink('dbname=userdb', '".$this->prefix."users')", $sql);

That would be roughly the alternative to the MySQL method above?

However, by the looks of that first dblink doc, the select statement for that table also needs to be within that dblink function, along the lines of:

$sql = str_replace($this->prefix."users", "dblink('dbname=userdb', 'select [userdb required columns here] from ".$this->prefix."users')", $sql);

Unless I've misinterpreted that dblink documentation, that wouldn't be feasible using that str_replace method above, would it? Or, have I completely misinterpreted that documentation, as is my usual forte?

Thanks again,

Matt

How would things be done for the likes of delete/insert/update? Just been thinking about that. (Dangerous thing for me, I know). But, the statement/action part couldn't be wildcarded or similar as such, could it?

Would it be simpler to just run a second instance of the dblayer, (adjusted accordingly), in parallel with the existing one specifically for the user/group/online tables, rather than trying to work this dblink into the existing dblayer class?

My minds obviously in thick mode again. The bit which I'm just totally not grasping is this:

dblink('dbname=userdb', 'SELECT * FROM ".$this->prefix."users')

With the SELECT statement being hardcoded in, how would it know what to do when a DELETE or such was part of the query, rather than SELECT?

Looks like annoying syntax when doing joins.

It's not because more is required than that. Check out the code examples given in the PostgreSQL documentation, you need to define each column individually.
I'm just going to say that there is no simple way to interact between databases in PostgreSQL