Seems that way. Careful testing will as always confirm the methodology
and implementation.
On 09/26/2016 06:29 PM, HP3 wrote:
Thank you!
So, the following is all true?
Recap'ing the solution (Jason's and yours combined):
1- On `after_flush`, call `history_meta.create_version()`
Agreed!
Thank you very much again!
--
You received this message because you are subscribed to the Google Groups
"sqlalchemy" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to sqlalchemy+unsubscr...@googlegroups.com.
To post to this group, send email to
Thank you!
So, the following is all true?
Recap'ing the solution (Jason's and yours combined):
1- On `after_flush`, call `history_meta.create_version()` for each object
that is changing so its `XYZHistory` instance is created and can be stored
within `session.info`.
2. On `before_commit`, do
On 09/26/2016 01:24 PM, HP3 wrote:
It seems like 0.9 does not have 'parent' in `SessionTransaction`.
Instead, it has `_parent`.
Would `_parent` suffice?
yes, use _parent. i pushed up docs to 1.0 and 1.1 which indicate that
_parent is the original attribute in the absense of .parent.
It seems like 0.9 does not have 'parent' in `SessionTransaction`. Instead,
it has `_parent`.
Would `_parent` suffice?
On Monday, September 26, 2016 at 12:09:55 PM UTC-5, HP3 wrote:
>
> Thank you Mike,
>
> So basically, I should move the cleanup from
> `after_commit`/`after_rollback` to
Thank you Mike,
So basically, I should move the cleanup from
`after_commit`/`after_rollback` to `after_transaction_end` and do it
**only** if `transaction.parent is None`?
Recap'ing the solution (Jason's and yours combined):
1- On `after_flush`, call `history_meta.create_version()` for each
On 09/23/2016 04:24 PM, HP3 wrote:
I just tried cleaning up at `after_transaction_end` but the whole
history stopped working.
It seems like `after_transaction_create`/`after_transaction_end`
surrounds the loading of a one-to-many relationship.Thus, when
`after_commit` happened, the
I just tried cleaning up at `after_transaction_end` but the whole history
stopped working.
It seems like `after_transaction_create`/`after_transaction_end` surrounds
the loading of a one-to-many relationship.Thus, when `after_commit`
happened, the `history_objects` dict had being emptied
Thank you Mike!
At this point, I am cleaning session.info["history-objects"] in
'after_commit' and 'after_rollback'. I was wondering about
'after_soft_rollback` too ...
(I discovered that history objects were being "re-added" when I was issuing
session.rollback() so I added the cleanup there
On 09/23/2016 10:18 AM, HP3 wrote:
Hello all,
Couple of weeks back (see [*]), while discussing that history_meta.py
performs an update and an insert for each session.dirty object whenever
session.flush() happens, Simon suggested the following solution to
coalesce all changes within the same
Hello all,
Couple of weeks back (see [*]), while discussing that history_meta.py
performs an update and an insert for each session.dirty object whenever
session.flush() happens, Simon suggested the following solution to coalesce
all changes within the same transaction into a single insert and
11 matches
Mail list logo
