Re: [dba-dev] postgres ODBC problem with view

2010-10-11 Thread Reizinger Zoltán
2010.10.11. 21:26 keltezéssel, Frank Schönheit írta: Hi Josh, If so, that's rather problematic. The CTID is a physical pointer ID, and as such can change out from under a current session if there are concurrent updates. Using the CTID to grab rows could result in errors or even seeing expire

Re: [dba-dev] postgres ODBC problem with view

2010-10-11 Thread Frank Schönheit
Hi Josh, > If so, that's rather problematic. The CTID is a physical pointer ID, > and as such can change out from under a current session if there are > concurrent updates. Using the CTID to grab rows could result in errors > or even seeing expired data. Which means SCOPE would be something oth

Re: [dba-dev] postgres ODBC problem with view

2010-10-11 Thread Josh Berkus
> Ocke knows more about this, but I suspect the PQ driver might deliver > that column's name as BestRowIdentifier, which might tempt our RowSet to > use it in some kind of "SELECT * FROM WHERE ctid=?" statement > (for re/fetching rows). If so, that's rather problematic. The CTID is a physical p

Re: [dba-dev] postgres ODBC problem with view

2010-10-11 Thread Frank Schönheit
Hi Josh/Zoltan, >> The data content could not be loaded. >> SQL Status: 42703 >> Error code: 7 >> ERROR: column "ctid" does not exist; > > That's very odd. ctid is a system column; no driver should be accessing > it, for any reason. Ocke knows more about this, but I suspect the PQ driver might d

Re: [dba-dev] postgres ODBC problem with view

2010-10-11 Thread Josh Berkus
Zoltan, > The data content could not be loaded. > SQL Status: 42703 > Error code: 7 > ERROR: column "ctid" does not exist; That's very odd. ctid is a system column; no driver should be accessing it, for any reason. -- -- Josh Berkus