On Apr 28, 2006, at 6:38 AM, Juergen Saar wrote:
No, that's the dark side of this kind of working.
For us this is really no problem,
for most installation we use informix without transaction
because it is much faster and we would need transactions
in about 1% of our software. So we have implemented
a kind of self-healing for most kinds of data-defects.
Not that I think it is a good solution in this case, but since you
have it in place already - here is how you can do transactions
wrapping multiple DataContexts (this will require 1.2 release)
http://objectstyle.org/confluence/display/CAYDOC/Understanding
+Transactions
Nevertheless it would be fine to have a fifo strategy for db-
statements in
cayenne.
I am yet convinced that this would be a good general DB commit
strategy (it might as well be, still I am not ready now to discuss
this in depth). But here is a thing - nested contexts do use the FIFO
strategy when committing from one to another, and the needed change
order information is available already during commit. So we'll gladly
accept a patch that implements a FIFO strategy as an option ;-)
Still I'd like to understand what causes the original problem... Is
this a SQLServer/Informix thing??? I've never seen it on Oracle or
Postgres.
Andrus
