Tracking changed fields on a Django model instance.
Makes a Mixin available that will give you the properties:
is_stale
stale_fields
As well as the methods:
save_stale()
Which will will selectively only update stale columns using the familiar
Model.objects.filter(pk=pk).update(**stale_fields)
pattern (but still
resolves F()
or auto_now
constructs).
Install the package using pip. Then use the instructions in "Using the Mixin in the Model".
$ pip install django-stalefields
or if you're interested in developing it, use virtualenv and
virtualenvwrapper. The default settings.py
will look for the
stalefields package in its current location.
$ mkvirtualenv django-stalefields
(django-stalefields)$ pip install -r example_app/requirements.pip
(django-stalefields)$ example_app/manage.py test testing_app
You need to set make two settings.py
tweaks:
# settings.py
INSTALLED_APPS = (
'stalefields', # must be first!
# the rest...
)
AUTO_STALE_FIELDS = True
This provides the methods and functionality automatically for all registered models.
from django.db import models
from stalefields import StaleFieldsMixin
class TestModel(StaleFieldsMixin, models.Model):
"""A simple test model to test stale fields mixin with"""
boolean = models.BooleanField(default=True)
characters = models.CharField(blank=True, max_length=80)
(django-stalefields)$ ./manage.py shell
>>> from testing_app.models import TestModel
>>> tm = TestModel(boolean=True, characters="testing")
>>> tm.save()
>>> tm.is_stale
False
>>> tm.stale_fields
()
>>> tm.boolean = False
>>> tm.is_stale
True
>>> tm.stale_fields
('boolean', )
>>> tm.characters = "have changed"
>>> tm.is_stale
True
>>> tm.stale_fields
('boolean', 'characters', )
>>> tm.save_dirty() # just saves the dirty fields via .update()
>>> tm.is_stale
False
>>> tm.get_stale_fields
()
Three reasons:
- Convenience
- Optimization
- Bug avoidance
When using signals, especially pressave, it is useful to be able to see what fields have changed or not. A signal could change its behaviour depending on whether a specific field has changed, whereas otherwise, you only could work on the event that the model's save() method had been called.
Any time you call boring old save()
inside Django, all columns are inserted
once again, which can be very heavy if, for example, you have lots of text in one
column or many indexes that don't need to be needlessly checked for updating. Only
updating changing columns via update()
is much faster, but requires lots of state
monitoring of your own accord. Put simply, this is nicer you your database!
Finally, if multiple threads call save()
for different operations, only the more
recent thread wins. If they both INSERTED only their column's changing values, that wouldn't
be an issue! These kinds of bugs are a nightmare to chase down...
This code has largely be adapted from what was made available at Stack Overflow and adapted from Forked from https://github.com/smn/django-dirtyfields and https://github.com/callowayproject/django-dirtyfields..