Обсуждение: New ODBC driver
What would be the best way to go about re-writing the ODBC driver to use libpq? (I haven't looked at the code in years, the last time I looked at it was to find a memory leak.) Is it possible to do it in stages? Is the transport sufficiently separated to do a quick change to use libpq as the transport, and then gradually use more and more? My only concern would be that a re-write would re-break a lot of functionality.
> -----Original Message----- > From: pgsql-odbc-owner@postgresql.org > [mailto:pgsql-odbc-owner@postgresql.org] On Behalf Of > markw@mohawksoft.com > Sent: 01 December 2004 13:50 > To: pgsql-odbc@postgresql.org > Subject: [ODBC] New ODBC driver > > What would be the best way to go about re-writing the ODBC > driver to use libpq? (I haven't looked at the code in years, > the last time I looked at it was to find a memory leak.) > > Is it possible to do it in stages? > > Is the transport sufficiently separated to do a quick change > to use libpq as the transport, and then gradually use more and more? > > My only concern would be that a re-write would re-break a lot > of functionality. I haven't looked at it in depth, but I think you will essentially just need to rewrite connection.c as a libpq wrapper. I don't think it would be realistic to mix the two approaches though - if only because they couldn't share a single database connection. Regards, Dave.
> > >> -----Original Message----- >> From: pgsql-odbc-owner@postgresql.org >> [mailto:pgsql-odbc-owner@postgresql.org] On Behalf Of >> markw@mohawksoft.com >> Sent: 01 December 2004 13:50 >> To: pgsql-odbc@postgresql.org >> Subject: [ODBC] New ODBC driver >> >> What would be the best way to go about re-writing the ODBC >> driver to use libpq? (I haven't looked at the code in years, >> the last time I looked at it was to find a memory leak.) >> >> Is it possible to do it in stages? >> >> Is the transport sufficiently separated to do a quick change >> to use libpq as the transport, and then gradually use more and more? >> >> My only concern would be that a re-write would re-break a lot >> of functionality. > > I haven't looked at it in depth, but I think you will essentially just > need to rewrite connection.c as a libpq wrapper. > > I don't think it would be realistic to mix the two approaches though - > if only because they couldn't share a single database connection. Why is this an issue? Why not create multiple connections? > > Regards, Dave.
> -----Original Message----- > From: markw@mohawksoft.com [mailto:markw@mohawksoft.com] > Sent: 01 December 2004 15:02 > To: Dave Page > Cc: pgsql-odbc@postgresql.org > Subject: Re: [ODBC] New ODBC driver > > > I don't think it would be realistic to mix the two > approaches though - > > if only because they couldn't share a single database connection. > > Why is this an issue? Why not create multiple connections? Efficiency for a start - but what would you use each for? Any application developer is going to expect everything to use the same connection - how do you direct a 'SET' statement? What about transaction isolation? Regards, Dave.
> > >> -----Original Message----- >> From: markw@mohawksoft.com [mailto:markw@mohawksoft.com] >> Sent: 01 December 2004 15:02 >> To: Dave Page >> Cc: pgsql-odbc@postgresql.org >> Subject: Re: [ODBC] New ODBC driver >> >> > I don't think it would be realistic to mix the two >> approaches though - >> > if only because they couldn't share a single database connection. >> >> Why is this an issue? Why not create multiple connections? > > Efficiency for a start - but what would you use each for? Any > application developer is going to expect everything to use the same > connection - how do you direct a 'SET' statement? What about transaction > isolation? > > Regards, Dave. OK, I'm going to take a look at it. I have a stupid question, where is the location of the ODBC source? Is it part of PostgreSQL? > > ---------------------------(end of broadcast)--------------------------- > TIP 7: don't forget to increase your free space map settings >
> -----Original Message----- > From: markw@mohawksoft.com [mailto:markw@mohawksoft.com] > Sent: 01 December 2004 16:21 > To: Dave Page > Cc: pgsql-odbc@postgresql.org > Subject: Re: [ODBC] New ODBC driver > > > > > > >> -----Original Message----- > >> From: markw@mohawksoft.com [mailto:markw@mohawksoft.com] > >> Sent: 01 December 2004 15:02 > >> To: Dave Page > >> Cc: pgsql-odbc@postgresql.org > >> Subject: Re: [ODBC] New ODBC driver > >> > >> > I don't think it would be realistic to mix the two > >> approaches though - > >> > if only because they couldn't share a single database connection. > >> > >> Why is this an issue? Why not create multiple connections? > > > > Efficiency for a start - but what would you use each for? Any > > application developer is going to expect everything to use the same > > connection - how do you direct a 'SET' statement? What about > > transaction isolation? > > > > Regards, Dave. > > OK, I'm going to take a look at it. > > I have a stupid question, where is the location of the ODBC > source? Is it part of PostgreSQL? > http://gborg.postgresql.org/project/psqlodbc, in the CVS section. Regards, Dave,