Re: Proposed change to disconnect in DBD::ODBC

2011-03-16 Thread Greg Sabino Mullane
-BEGIN PGP SIGNED MESSAGE- Hash: RIPEMD160 > This area is already inconsistent. From DBI under disconnect: > > "The transaction behaviour of the disconnect method is, sadly, > undefined. Some database systems (such as Oracle and Ingres) > will automatically commit any outstanding chang

Re: Proposed change to disconnect in DBD::ODBC

2011-03-16 Thread David Nicol
A non-fatal warning is the right thing here. do nothing |warn |die notifies in new dev no | yes | yes darkpan safeyes | probably |no -- "This is not a 'bug'. Being

Re: Proposed change to disconnect in DBD::ODBC

2011-03-16 Thread Martin J. Evans
On 16/03/11 16:07, Greg Sabino Mullane wrote: -BEGIN PGP SIGNED MESSAGE- Hash: RIPEMD160 but in the second case DBD::ODBC will rollback the transaction before disconnecting and there is no error or warning. I don't really like this behaviour and would like to change it so in the second

Re: Proposed change to disconnect in DBD::ODBC

2011-03-16 Thread Greg Sabino Mullane
-BEGIN PGP SIGNED MESSAGE- Hash: RIPEMD160 > but in the second case DBD::ODBC will rollback the transaction before > disconnecting and there is no error or warning. > I don't really like this behaviour and would like to change it so in the > second > case above, DBD::ODBC issues a war

Proposed change to disconnect in DBD::ODBC

2011-03-16 Thread Martin J. Evans
DBI does not define what happens wrt transactions when disconnect is called. Up to DBD::ODBC 1.29, DBD::ODBC always called SQLEndTran(SQL_ROLLBACK) to roll back any outstanding transactions (if AutoCommit was off) before calling SQLDisconnect. This means the outstanding transaction that is roll