Accessing many-to-many relations before saving is no longer allowed.
Reorder all operations consistently:
1. Validations
2. Modify own fields
3. Perform save by calling super().save()
4. Modify related objects and clear caches
Especially clearing caches should be done after actually saving, otherwise the old data can be
re-added immediately by another request before the new data is written.
Previously, triggers lived only in a particular migration file. With
this change, code for the triggers resides in the model, and their
lifecycle is managed through normal Django migrations.
Authors have Websites. that might be interesting for BookWyrm users. Now there is a field specify the website of an author and it is shown tu users under the wikipedia link
Adds max char for ISFDB ID (6) and ISNI (19 - 16 plus three -)
This should be enough caching when renderering pages like the feed, but
not so much that editing a book or author will not show the updated data.
At least without having to do some clever cache busting.
* New ID: Audible ASIN
Audible belongs to Amazon BUT they do not share the same IDs. The Audible ASIN of an audiobook is never the same as the Amazon ASIN.
Yeah, I know, Amazon is great. The fact that the ASIN is a good distinction for different works and editions bothers me more than I will ever be willing to admint.
* New ID "ISFDB"
Internet Speculative Ficiton Database ID for books and authors.
Links to the entry if set.
* Added aasin to test
Added aasin to test
* the answer expects more emptxy fields...
It feels janky to remove a more granular name designation, but all these
first/last name fields were algorithmically populated by a dubious
process of splitting the name by a space character. If it makes sense to
have first/last name fields, it should be re-added with some
consideration.
