| Age | Commit message (Collapse) | Author |
|
|
|
|
|
Regression in 2f9861d823620da7ecb291a8f005f53da12b1e89.
|
|
|
|
Thanks Tim Graham for the review.
|
|
|
|
|
|
|
|
|
|
|
|
Thanks Tim Graham and Simon Charette for the reviews.
|
|
|
|
|
|
Per deprecation timeline.
|
|
|
|
|
|
|
|
These values are duplicated in the super().
|
|
* Avoid some unnecessary attribute lookups, e.g. access signals directly rather than from module
* Alias some repeat accesses inside the method to use the slightly faster local lookups
* Use tuple to iterate remaining kwargs as it's faster to construct
* Cache Field.get_default() to avoid running through all the logic on every call
* Use a cached list of the properties on the model class to avoid repeat isinstance() calls
|
|
|
|
|
|
|
|
|
|
|
|
The smart_* version should only be used when a lazy string should keep
its lazy status.
|
|
|
|
Previously, empty values were saved as strings.
|
|
|
|
Thanks Tim Graham for completing the initial patch.
|
|
|
|
deferred instance loading.
|
|
|
|
|
|
form/model fields.
|
|
The only reason why GenericForeignKey and GenericRelation are stored
separately inside _meta is that they need to be cloned for every model
subclass, but that's not true for any other virtual field. Actually,
it's only true for GenericRelation.
|
|
|
|
|
|
|
|
Thanks Common Code for financing the work on this commit.
|
|
Thanks Tim Graham for the review.
|
|
|
|
|
|
|
|
Thanks Simon Charette for the test.
|
|
|
|
|
|
|
|
This reverts commit 71ebcb85b931f43865df5b322b2cf06d3da23f69.
|
|
|
|
|