On Wed, May 04, 2011 at 07:03:31PM +0200, Marek Więckowski wrote:
> (and this is why I was looking into this in the first place). There is a
> danger that client programs will continue issuing queries while believing that
> they are in a transaction... They do expect db errors and rolled back
> transactions, but not that their begin-commit section would be executed only
> partially.
I don't understand. If they are prepared for errors, then if they
have any error they have to roll back to the beginning of the
savepoint or the transaction if there's no savepoint. What is this
"partial execution" of which you speak? Nothing is partially
executed: if the transaction rolls back, everything is lost.
> good (and the alternative sounds more complex: it would require exposing extra
> info to the programs using this library, and add handling of reconnect
> situation in each of these programs etc.).
[. . .]
> 4. if we were in a trans before disconnect, then immediately after
> reconnecting we would create a trans-in-error.
You're going to have to expose trans-in-error to the client anyway.
What's the difference between that and exposing the program to needing
to fix its connection?
A
--
Andrew Sullivan
ajs@crankycanuck.ca