I'll try to write a ready-for-checking patch ASAP. It is easier to
discuss about concrete and full patch.
- Anssi
On 08/28/2013 11:44 AM, Marc Tamlyn wrote:
I think we should implement `select_on_save`. It gives the easiest way
for someone to revert the feature if they need to. I know we're adding
another Meta flag, but I'm not that concerned about that.
On 28 August 2013 09:23, Anssi Kääriäinen <[email protected]
<mailto:[email protected]>> wrote:
I just created a new ticket
(https://code.djangoproject.com/ticket/20988), and marked it as
release blocker for 1.6.
The summary of the issue is that there are some cases where the
new save() algorithm will not save existing objects (for details
about the algorithm, see #20988). This happens when an UPDATE of
an existing row returns zero matched rows. At least PostgreSQL ON
UPDATE triggers returning NULL can cause this condition. If an
user is hit by this issue, then currently the options are to
remove the trigger, or to use raw SQL for update of existing
objects. There might be 3rd party backends or other conditions
where the only option is raw SQL.
It would be nice to have better alternatives than raw SQL. Some
options:
- Change back to old algorithm.
- Change the Model._do_update() to include forced_update flag.
If this is done, it will be possible to override to _do_update()
to use the old style algorithm. Unfortunately _do_update() is
private API, but using that is likely better for most users than
raw SQL.
- Add a select_on_save flag now, that is before release. The
flag effectively forces use of the old algorithm. The
implementation is straightforward, see
https://github.com/akaariai/django/commit/1ca398b0352b11868ec236f92cf17b6ce82ff88c.
Docs missing, but otherwise ready.
- Wait and see. If there are enough complaints, add
select_on_save or some other workaround later on (that is, mid
1.6, or in 1.7).
In addition, better release notes about this issue are needed.
I think reverting to the old algorithm isn't a good option. The
new algorithm will save one query in most cases. I will do the
_do_update() signature change now, it is good for those who want
to use that particular private API in any case.
The real question is should something be done about select_on_save
flag now?
- Anssi
--
You received this message because you are subscribed to the Google
Groups "Django developers" group.
To unsubscribe from this group and stop receiving emails from it,
send an email to [email protected]
<mailto:django-developers%[email protected]>.
To post to this group, send email to
[email protected]
<mailto:[email protected]>.
Visit this group at http://groups.google.com/group/django-developers.
For more options, visit https://groups.google.com/groups/opt_out.
--
You received this message because you are subscribed to the Google
Groups "Django developers" group.
To unsubscribe from this group and stop receiving emails from it, send
an email to [email protected].
To post to this group, send email to [email protected].
Visit this group at http://groups.google.com/group/django-developers.
For more options, visit https://groups.google.com/groups/opt_out.
--
You received this message because you are subscribed to the Google Groups "Django
developers" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
To post to this group, send email to [email protected].
Visit this group at http://groups.google.com/group/django-developers.
For more options, visit https://groups.google.com/groups/opt_out.
