Bruno CROS wrote:
Looking into OJB implemenations, it seems that deletePersistent use the
transaction of the current thread. So ok for the SQL transaction sequence,
sorry for asking in last mail. I was confused between
DB.deletePersistent(o),
ExtTx.markDelete(o), and ExtTx.deletePersistent(o) methods. It seems all do
the same.
They are nearly the same (see my previous post).
I can't get rid of those dead lock, occurring when update/delete object
in complex and different transactions.
Do i have to lock objects to be deleted to avoid concurrent updating
transaction (plausibly locking DB records) ?
If you use DB.deletePersistent(o) the object will be immediately (write)
locked, thus it couldn't happen that another tx locks the same object
too (ExtTx.markDelete(o) doesn't write lock the object itself, thus you
first have to write lock it).
regards,
Armin
Does someone use such a technique ?
Thanks for help.
On 12/8/06, Bruno CROS <[EMAIL PROTECTED]> wrote:
Hi,
Just one question.
Using ODMG, if deletePersistent(o) is called from the Impl.getDatabase(),
how "SQL DELETE" query post can be done with the other
"INSERT/UPDATE" in
the same SQL transaction without refering THE transaction? Why ODMG
delete
object tutorial talk about deletePersitent way, whereas
ExtTx.markDelete(o) would
be more appropriate considering SQL queries sequence in transaction?
Need help.
Regards.
Bruno
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
