> Isn't the 'try' statement rather similar to a 'savepoint' command? I
> realize it would be difficult to override the behaviour of try {...}
> catch (...) {...}, but it shouldn't be too hard to wrap it somehow for
> exceptions in database code.
Yes, but I believe the OP was getting two levels of his application
mixed up: he was doing something that caused a rollback in the
*database*, then hoping to recover in a catch block in the
*application* without terminating the aborted transaction in the
database. Or so I gather.
You are right in another post about the purpose and design of
transactions, and don't use the discussion here as a model, though
drivers seem to often do weird stuff with transactions behind your
back. Psycopg (python) does an implicit begin, so you must commit,
which then starts another begin automatically. I think you can set a
handle to do autocommit, but I never do. This seems best because it
forces you to handle transactions explicitly, but I can imagine other
(bad) approaches, and Spring may use them (though I think the
difficulty is that MS-SQL is sloppy, not Spring, and that the OP is
still getting used to TX's and MS-SQL covered up some things that
shouldn't have been covered).
W
