Good pointers Sergi and Dario.
So my checklist is:
- test large transaction from memory point of view
- think about how to identify rows to be the same row
- do more investigations before coding
- Expect weird and disappointing things to come your way
With that and Thomas' long instructions on providing patches
I will move forward. I will first just do something simple and stupid
and try it in my own project just to see if I can make it work at all.
I will keep you posted.
- rami
25.11.2010 23:50, Sergi Vladykin kirjoitti:
So if someone would like to move the constraint check to the end of
transaction
one would need to create some kind of changeset caused by the transaction.
The simple way to do it would be
List<Row> oldRows, List<Row> newRows
that would simply be a list of rows that have been somehow affected by
the transaction.
These lists could be populated in the same place where the constraints
are now being checked.
And then at commit time the constraints could be actually checked.
On long transactions or large data imports this can cause out of
memory errors. Also the same row can be modified mutiple times in
transaction so in these lists you will hold multiple versions of the
same row. This is only simple problems I see but I think there will be
more weird ones. Anyways just keep all changes in memory obviously is
not very good idea.
In my opinion here is needed to store only failed row keys, manage
them (delete key if row was deleted from table or if this row was
updated and constraint was satisfied, may be more complex things will
be needed for referential constraints) and if on commit at least one
failed row key is still here then transaction should be rolled back
and exception thrown.
And for all you speed freaks this could actually increase speed because
now the constraints
are checked twice for every statement (before and after the statement).
Doing it at transaction boundary
would require running them only once :-)
It is not true. Before statement called one part of constraints and
after another part but not all constraints two times (see isBefore()).
Comments?
Should I give this a shot?
I think it would be nice if you will implement this feature but please
do more investigations before coding.
regards,
Sergi
--
You received this message because you are subscribed to the Google Groups "H2
Database" group.
To post to this group, send email to [email protected].
To unsubscribe from this group, send email to
[email protected].
For more options, visit this group at
http://groups.google.com/group/h2-database?hl=en.
