I've re-worked the workflow form endpoint, so that it only accepts the data that should be in the form, and ignores any other values that come back from the front end. It seems Formly has some bugs that were introducing confusing information, and I want everything to behave consistently.
I had to re-work some of the tests, which were relying on an ability to set data through a form post without having a corresponding form to do so.
To do so, I changed the WorkflowProcessor's reset method to be static, it will try to instantiate the current workflow and execute a cancel_notify event, but if that fails, it will continue and always return a new workflow processor using the latest spec.
While working on this ticket I found a logic error in the test. I expected both default_value and value_expression. It turns out the default_value is always there, it's just null if we don't set it.
Modified the existing test for hidden and required, and added tests to make sure my passing logic is now correct.
I added a test in tests.test_email_script to make sure we get the default email_address.
I also found and fixed a bug in workflow_service.populate_form_with_random_data where we would overwrite default_values with random text because of a missing continue.
(We don't want to continue if we have repeating fields though.)
Added a human readable name to created user
added last updated task time and name,
added is_review on the workflow which is triggered by a swimlane with a name present.
Notably - it does NOT update or change any of the performance concerns that we were having with this.
- If the default was False, we failed an early test and returned None. We now only return None if default is None
- We have a better test for deciding if the default value we return should be True or False. We used to return True if the value was the string 'false'.
In api.study.update_study we test the study status and call the new WorkflowService method process_workflows_for_cancels.
In services.workflow_service we added the new method process_workflows_for_cancels. It loops through workflows for a study, and resets them if they are in progress.
In services.workflow_processor, we changed the reset method to be an instance method so we can call self.cancel_notify.
In tests.test_lookup_service we changed the call to WorkflowProcessor.reset to reflect the change from class method to instance method
Fixing workflow sync paths that were incorrect.
Repairing a suddenly failing test in files, that just don't make no sense.
Bumping spiffworkflow that contains a fix for issue #155
2. Disabling the token timeout for now, to see if this corrects the issues Alex is having with lost work.
3. Raising more thoughtful error messages for unknown lookup options.
4. Providing better validation of default values and injecting the correct value for defaults related to enum lists of all types.
5. Bumping Spiffworkflow library which contains some better error messages and checks.
Making all the lookup field names consistent.
Fixing the lookup service which was failing at times trying to find the correct field to use for building the lookup table.
Updating validation to check for additional fields and properties.
When connexion level errors occur, wrapping it in an API Error to be consistent.
Add the lane information to the Task model.
Drop the foreign key constraint on the user_uid in the task log, as we might create tasks for users before they ever log into the system.
Add a new endpoint to the API called task events. It should be possible to query this and get a list of all tasks that need a users attention.
The task events returned include detailed information about the workflow and study as sub-models
Rename all the actions in event log to things that are easier to pass over the api as arguments, make this backwards compatible, updating existing names in the database via the migration.
Throughly test the navigation and task details as control of the workflow is passed between two lanes.
Adding checks in the API to assure the correct person is completeing a task based on the task's lane.
Adding lane to the Navigation item.
Adding a check to assure that unique user ids can be identified using task.data
Added some additional ldap entries to make testing and development easier.
Removed a big chunk of duplicate code from task_tests_api
Modified some of the helper functions to make it easier to test as specific users.
Added some additional bpmn models to the tests for testing lanes and roles.
* The Task.title returned to the front end will now attempt to process the "display_name" property for dot-notation syntax, making it possible to use this for multi-instance tasks,
but will work in all cases where we want he title to change based on values in the data model.
* Fixing a bug the test_study_api where it wasn't updated when we made recent changes to the different states of a study.
*
In order to allow proper deletion of tasks, we no longer merge data returned from the front end, we set it directly as the task_data.
When returning data to the front end, we take any previous form submission and merge it into the current task data, allowing users to keep their previous submissions.
There is now an "extract_form_data" method that does it's best job to calculate what form data might have changed from the front end.
* TaskEvents now contain the data for each event as it was when the task was completed.
* When loading a task for the front end, if the task was completed previously, we take that data, and overwrite it with the lastest data, allowing users to see previously entered values.
* Pulling in the Admin branch, as there are changes in that branch that are critical to seeing what is happening when we do this thing.
* Moved code for converting a workflow to an API ready data stricture into the Workflow service where it belongs, and out of the API.
* Hard resets just convert to using the latest spec, they don't try to keep the data from the last task. There is a better way.
* Moving to a previous task does not attept to keep the data from the last completed task.
* Added a function that will fix all the existing RRT data by adding critical data into the TaskEvent model. This can be called with from the flask command line tool.
mike cullerton
Dan Funk
Kelly McDonald
Aaron Louie
Carlos Lopez