Release Notes v6.8

What's new?

• CONVERT TEXT DURATION TO FLOAT pattern. This pattern casts a text field representing a time duration into a possibly-fractional number of seconds/minutes/etc.

What improvements have we made to existing features?

• Ad-hoc derived fields will be immediately pinned when created in Case Review.
• Support "Prevent Passive ELT" on Measure Objects.
• Improved the workflow around updating saved ELTs. Now, when you load a saved ELT you'll be given the option to either update it, or save it as a new ELT.
• Added Saved ELT Trigger Endpoints panel to the Integration Manager, moving this information out of the Saved ELT panel in Object Workshop.
• The "Run affected reports" and "Email when done" checkboxes in the ELT Preview panel have been replaced by icon buttons that toggle the same settings.
• Allow explicit cast to date or datetime for DATE OFFSET pattern. By default the data type will be datetime if the added interval is minutes or hours, and date otherwise.
• Added an option in the developer panel in Measure Workshop and Object Workshop to view the query plan and cost of a measure or object of RDBMSes which support EXPLAIN queries.
• When range imports are activated for import objects, the range import will normally iteratively import every file, starting with the chosen filename and ending with the range end filename, in alphabetical order. However, in the circumstance where there are filenames that are between the start of the range and the end of the range which must be skipped, those files can now be screened out by setting a case-sensitive secondary filter for the range. A range import preview has also been added to show to the user which files will be imported, based on the selected criteria.
• When setting up import objects to impute the data effective date from the filename and to use a mask to dynamically choose the most recent filename in the container matching that masked filename, users might run into the problem that there are characters in the filename which are not part of the date mask but which should nevertheless be allowed to vary from filename to filename while still being considered "matches". An example of this could be if the filenames have not just the date in the filename but a time value as well; these time values should be effectively ignored. To perform these imports, users are now able to enter the character * (asterisk) as a single-character wildcard. For example, they can type my_file_YYYYMMDD******.txt as the storage filename. To facilitate this typing, a clipboard icon has been added alongside the storage container name, which will copy the current filename into the clipboard. Users can then immediately paste the filename into the dropdown and replace the variable characters with asterisk.
• Support AWS SageMaker for Bespoke Model Objects. SageMaker functionality can be turned on by adding the SAGEMAKER_IAM_ROLE_ARN environment variable to Ursa Studio.
• Redshift implementations now support an optional USE_REDSHIFT_DISTKEY environment variable. If this variable is set, every non-precursor, non-view object will use its patient ID field as the table distkey if the patient ID field is set in the object metadata.

What's been fixed?

• For PostgreSQL and Redshift databases, updated the SQL for the ELAPSED TIME pattern to avoid an intermediate resolution of inputs to the nearest day, which fixes rounding errors when the inputs are not dates but datetime data types.