* add cache_dir option to OpenVINO EP
* update provider options test to include cache_dir
* use forward slash instead of string concatenation
* fix cache_dir placement in provider options assertion
* Added SVG Support
* Removed comment
* Server Test fixes
* Sorted the mimetypes
* Reverted mimetypes.assettype
* Lint
* fix test
---------
Co-authored-by: Alex Tran <alex.tran1502@gmail.com>
* fix(web): keyboard shortcut handling
* drop executeShortcuts in favor of action
* fix merge
---------
Co-authored-by: Jason Rasmussen <jrasm91@gmail.com>
* Fixes remote full / thumbnail provider
* Adds image cache manager to both remote image providers
format
format
Fix typo in equals
remove unused import
renames image loader
* Adds height and width to the image cache for thumbs
format
* Uses a separate remote and thumbnail cache
format
* Fixes key name
* Changes uri to string, fixes comment
* Chunk events are optional and remote thumbnails don't report chunk events
* better exception handling
---------
Co-authored-by: Alex Tran <alex.tran1502@gmail.com>
* feat(mobile): Adds file upload progress stats: current upload file size uploaded, current file size and formatted bytes per second upload speed. Closes#7379
* chore(mobile): Fix stan issues
* chore(mobile): Remove non-'en-US' translations, as I saw on another PR review (just looking around) that localisation is done via Localizely and this was the instruction (to only provide the en-US localisation).
* fix(mobile): Provide boundary checks to ensure overflow issues are accounted for on erroneous upload speed calculation, sometimes the numbers received back from the upload handler can be a bit wild.
* fix(mobile): Some heuristic bug fixing. Whilst thinking what could trigger overflow issues or 'zero' readouts, left over values from the previous file may do that. So adding the last upload sent bytes to the values to be reset may help! The time isn't necessary, as the period/cycle is inconsequential in this circumstance, well it should be anyway.
* fix(mobile): Actually, in combination to the last commit, some more heuristic bug fixing. I was thinking it would be advantageous not to reset the update time, as it would trigger a quicker first upload speed calculation. However, I realised that could also cause the calculation to be incorrect on the first cycle as the period wouldn't align. Not really sure if it would be a big deal, but I'm taking wild guesses in the dark here. Again, some purely heuristic debugging as I can't re-produce the underlying issue. This is mainly just ensuring that the state is fully reset and is a known state at the beginning of each file as a common strategy to reduce issues.
* refactor(mobile): Move the UI for the file progress to underneath the progress bar, it makes more sense there than in the file information table which contains only static information pertaining to the file itself. Switching to a monospace font to keep the UI from jumping around as the numbers change.
* refactor(mobile): In order to have the UI always present an 'active' upload speed (as per the discussion on PR #7760), this stores the 'upload speeds' (capped at the latest 10) in a list and calculates the current upload speed as the average over them. This way the UI can always display a 'constant' upload speed during uploading, instead of starting a fresh when each file starts uploading. Limiting it to the 10 latest keeps the average somewhat recent and ensures some level of sensible memory allocation.
* clean up usage
* i'm not updating all these tests
* update tests
* add indices
* add indices to entities
remove index from person entity
add to face entity
fix
* simplify query
* update sql
* missing await
* remove synchronize false
