Subject | RE: [IB-Java] Re: Dying connections with interclient |
---|---|
Author | Jim Starkey |
Post date | 2001-04-18T20:24:02Z |
At 09:47 AM 4/18/01 +0200, Goedhart, Andrew wrote:
is direct access to the Firebird server on 3050. The other is a
bridge between JDBC and Firebird semantics.
Direct access to 3050 is quit doable. The protocol, although based
on XDR, is a quite simple request/response protocol. Other than
a little work staging blocks of records, no significant problem
at all.
Mapping JDBC semantics into Firebird is more work, mostly on the
metadata side. There just isn't a simple 1:1 mapping between
the many JDBC/ODBC mandated metadata result sets and the OSRI
system relations. A full implementation would probably require
some intelligent staging of meta-data within JDBC "driver".
On the other hand, much of the code could be recycled from the
existing interserver until the Interbase Public License. The
trick is staging enough (but not too much) information to implement
the database meta data calls. Or even better, if somebody could
pry loose the lower layer of the IBPhoenix ODBC driver the skin-
flinty old geek sitting on the code to nobody's advantage, you'd
be way ahead of the game.
My guess (which Ann doubles if not quadruples) is a guru-month
for a completely usable driver. Full JDBC compliance could easily
take another three, four months, but who really cares?
Jim Starkey
> Does this mean that there is a JDBC driver available that doesn't useThere are two quite different problems that need to be solved. One
> Interserver that can act as a starting point for a Firebird open source
> implementation of java interbase connectivity. Or am I just being hopefull
> :-)
>
> If this is the case where can I get hold of it and a description of the port
> 3050 protocol so that things can begin.
>
is direct access to the Firebird server on 3050. The other is a
bridge between JDBC and Firebird semantics.
Direct access to 3050 is quit doable. The protocol, although based
on XDR, is a quite simple request/response protocol. Other than
a little work staging blocks of records, no significant problem
at all.
Mapping JDBC semantics into Firebird is more work, mostly on the
metadata side. There just isn't a simple 1:1 mapping between
the many JDBC/ODBC mandated metadata result sets and the OSRI
system relations. A full implementation would probably require
some intelligent staging of meta-data within JDBC "driver".
On the other hand, much of the code could be recycled from the
existing interserver until the Interbase Public License. The
trick is staging enough (but not too much) information to implement
the database meta data calls. Or even better, if somebody could
pry loose the lower layer of the IBPhoenix ODBC driver the skin-
flinty old geek sitting on the code to nobody's advantage, you'd
be way ahead of the game.
My guess (which Ann doubles if not quadruples) is a guru-month
for a completely usable driver. Full JDBC compliance could easily
take another three, four months, but who really cares?
Jim Starkey