Document versioning
Store many versions of the same document, download or revert to a previous version. Each version can have its own separate access control too.
Released: April 29, 2022
Status: Stable
Trashed documents are now filtered from the cabinet document retrieval methods. This brings code parity with tag documents which work in a very similar way.
The LUT values used when masking an asset for pasting via Pillow's point() are now typecasted to integer which is what the function expects.
Updated the Debian Docker image from version 11.2-slim to 11.3-slim.
Downgraded the Python Docker image from version 3.11-slim to 3.10-slim.
Pin Jinja2 version to workaround Sphinx bug. Sphinx Jinja2 dependency is not pinned or immutable, and causes the installation of an incompatible version breaking builds.
Fixes the paths to the OTP backends shown in the documentation.
Added community contributed Portainer installation files and documentation chapter.
Improved the document metadata edit form validation.
The disabled attribute was removed from the metadata type label field to avoid having its value removed after a refresh when there is a validation error.
Removed the required flag from the value field when there is a required metadata for a document. The previous behavior cause the tabular form to display "(required)" in column title confusing users and causing them to think that all metadata type fields were required.
Validation errors are now raised for specific required metadata entries and not for the entire form. This help users better understand which metadata field needs to be corrected.
Improved the required metadata validation logic to take into account existing values and empty forms when data was entered into the field but the update checkbox was left unchecked.
Improved Python 3.10 compatibility by adding a compatibility module to encapsulate import of the Iterable class.
Removed the hardcoded search model variable name from the search_box.html template. The variable name is now extracted via a template tag to ensure backward compatibility even if the variable name is updated again in the future.
Fixed the search model API URL reference by having the API and view use the same variable to reference the search model.
Used the SEARCH_MODEL_NAME_KWARG instead of hard coding the search model API URL reference.
Retry failed bulk indexing tasks.
Add max retry value to task_index_search_models.
Added improved search tasks error logging.
Trashed documents are now filtered from the tag document count column.
