Changelog

Version 2.1.5

Add the canvas_identifier option in configuration to allow customizing the identifier for submissions

New clean command added to REPL which removes all marking information (candidates/ and marksheet.yml)
Fix off-by-one error in progress tracker
Change default shell for running commands to /bin/bash

Ignore hidden files when performing os.listdir()
Add running time per test to results
Can now set submission_comment in configuration (for Canvas) to add some text comments along with the posted report.
Can change the import_command used to import files in prepare to soft/hard link them instead of copying, etc.
Can now use marker.getLMSSubmissionURL() to get (for Canvas) the URL to open the submission directly.

Can now download all files in Canvas submission by omitting file_name in configuration.

API largely the same, but internals restructured to make it very simple to write wrappers over it (such as a web-server).
Use of Abstract Base Classes for Console and LMS objects, to make extended implementations have necessary methods.
Test results are now stored in a parseable JSON format instead of directly being dumped to report
Report generation now uses Jinja based on the results, much easier to customize.
Tokens, CLI history, default templates / configurations now all stored in ~/.config/marker/

Redesigned internals, can now import an instance of the Marker class in your Python code and script testing / handling of marks, etc. (Documentation for this is not available yet, but you can look in src/repl.py for code examples)
Now uses Python’s async abilities instead of multithreading to speed up API requests
Interactive CLI now available, with autocompletion of commands / student IDs.
Ability to run each command for individual students through the CLI directly
Can see some basic statistics from the marksheet in the CLI directly