Release v6.7.2 (2023-11-03)

Published 2023-11-03 10:47:00Z

• fylr_checksums.txt

• fylr_v6.7.2_darwin_amd64.tar.gz

• fylr_v6.7.2_darwin_arm64.tar.gz

• fylr_v6.7.2_linux_amd64.tar.gz

• fylr_v6.7.2_windows_amd64.zip

This is release is a bugfix release and it is recommended for all users. The update does not require a re-index.

Fixed

• /api/schema, /api/mask: Retrieving an old version of a data model (not CURRENT or HEAD) would use the internal id instead of the actual version of the model to retrieve data. In some databases the model ids have gaps, so that this bug shows during a full initiale synchronisation of an object store. This fix also adds more checks and indexes to newly created databases for consistent model storage.

• Base config load & save of locations: This bug showed on all instances after the update to 6.7.0 and resulted in lost default storage configurations. After setting and saving the storage locations again in the base config, the fixed bug showed only after locations where renamed using the location manager. The bug fix makes sure that renamed locations are handled properly and not losing their settings in default location settings again.

• Windows: Fixed sending of emails. The compilation of emails was broken since we introduced the embedded resources in 6.7.0. Fylr which still used the disk based resources where not affected by this bug.

• Concurrent access to /api/config with the file worker queue not fully done, could result in a full panic and crash of fylr. The panic says concurrent map writes. This was fixed by protecting the access to GET /api/config using a mutex.

• Custom Logo in Emails was fixed. This patch brings the proper template to show the custom instance logo in emails.

• Login: During login, explain a non privileged user that he cannot login due to licensing issues. Before this problem was hidden behind a generic login error.

• Export in CSV format: Fixed exporting of _system_object_id_parent and _tags.

• /api/suggest: Don't include delete objects (for fields and linked objects suggestions) in the results.

• /api/preset: Fixed permission check for users with the system.rightpresetmanager right to be able to save presets.

• Collection sharing: Some share links required a double login in order to access the collection. This was fixed, only one login is required now.

• Loading and indexing of collections could be broken if a previously configured create object setting which uses default tags was deleted. After that, the loader would panic and not be able index or load affected collections any more.

• Colorprofiles are now properly stored in the image metadata if --strip is used. That is the default in our standard recipes. So now, the default profile sRGB IEC61966-2.1 is mentioned in the metadata.

• Fixed formula columns plugin (and other) when used in object types with the new data type geo_json.

• fylr restore: Support custom payloads where no technical_metadata is set in the JSON payload. Before, restore would panic.

• also see Frontend, below

Improved

• If during startup, configurations are misconfigured in fylr.yml, a proper error instead of a panic is shown.

• Synchronisation of files during the file worker job sync was sped up by adding an index to the database. This problem most prominently showed in instances with more than 100.000 files.

• fylr restore: Improved generation of file names for uploads. If the JSON payload defines an extension and the filename has a none or another, the JSON defined extension is added to the filename.

Frontend

• The user CSV importer has been updated and now allows importing passwords from a CSV.

• The search bar now displays more clearly when a wildcard is being used in the autocomplete popup.

• Metadata mapping for image series within nested fields has been improved.

• #69864: Linked object filters have been fixed to work correctly with hierarchy object filters.

• The shared collections panel has been improved, and pagination has been added. If there are more than 100 collections in this panel, they will now be paginated. This solves a performance issue on instances with thousands of collections.

• Number fields in the CSV importer have been corrected. They were not imported correctly when multiple numbers were included in the same cell.

• Improved the behavior of date range fields using text representation in the CSV importer.

• Fixed an error when loading the transitions panel.

• The linked objects panel in quick access has been improved to support linked fields configured in the data model as expert search or filter. Previously, only those with expert search enabled were displayed.

• Added support in the CSV importer for finding tags by shortname and reference.

• The detail sidebar has been improved. Now, the mask used for each object type is remembered. Previously, if you changed the object type, the saved mask was lost, and the default mask was used.

• The filters section in the CSV importer has been improved, allowing more options for filtering rows.

• Fixed an error in the badge for linked objects in the search input.

• A new option has been added to the system fields configuration table in the mask editor to hide all system fields that are not represented in other rows of the table. This allows creating masks that completely hide the footer, for example.

• The "Select language for each field in editor" option in the language panel now also affects system multilanguage fields.

• Fixed searches in quick access (like Created today) now use the "Resources" filter like any other search.

• Fixed GeoJsonColumn editor input when there is no GeoJson Plugin installed on the instance. Now it will show a JSON editor.