On Apr 24, 12:12 am, Baz wrote:
> What was the conflict?
As I said before, there is code that does not run in a transaction,
but that should.
Here is what I assume causes the problem. There are two functions, f()
and g(). f() modifies the data store in a consistent way, using a
transaction. g()
What was the conflict?
On Fri, Apr 23, 2010 at 3:01 PM, Felix E. Klee wrote:
> Solved! Looks like I found the error: That possibly conflicting code
> does *not* run in a transaction, and it should.
>
> I'm glad that it's indeed me who is the idiot. ;-)
>
> --
> You received this message because
Solved! Looks like I found the error: That possibly conflicting code
does *not* run in a transaction, and it should.
I'm glad that it's indeed me who is the idiot. ;-)
--
You received this message because you are subscribed to the Google Groups
"Google App Engine" group.
To post to this group,
On Apr 23, 11:51 pm, "Felix E. Klee" wrote:
> since any conflicting
> transaction can only be one that reads from the data store.
Sorry, that's wrong. There could be a conflicting transaction that
writes to the data store.
Still, I have no clue why the parent version number is decreased or
rever
I found something interesting:
1. The transaction is indeed atomic, as expected. So it does add the
child and increase the parent version number. I saw this in the logs:
I had added debug output that shows the version number immediately
after the transaction finished.
2. However, at some point so
> Do you fetch the self.data_version entity inside the transaction or
> outside of it?
Inside, of course. The code I posted is run as part of the transaction.
--
You received this message because you are subscribed to the Google Groups
"Google App Engine" group.
To post to this group, send emai
Felix E. Klee wrote:
On Apr 23, 4:13 pm, Andi Albrecht
wrote:
Are you sure that you're not working on a parent model, that's somehow
cached and therefore out-dated when you start one of those
"non-atomic" transactions?
That's the code that's used to increase the parent version:
sel
On Apr 23, 4:13 pm, Andi Albrecht
wrote:
> Are you sure that you're not working on a parent model, that's somehow
> cached and therefore out-dated when you start one of those
> "non-atomic" transactions?
That's the code that's used to increase the parent version:
self.data_version = str(int(se