You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
One route is to say "No, this is not a feature/goal of DOJO. We will export to Canvas/other LMS, good luck!". But then you need a LMS, and historically we haven't reached for tools like Canvas because how hard is it to implement grading and deadlines? Probably this is not the path forward.
The issue of "arbitrary grading schemes" is that it's roughly equivalent to "arbitrary code". We could play sandbox games, and such, but ultimately when a student clicks on their grade page, they'd probably like to see the answer in <500ms. Sandbox games sound slow / a decent amount of code/infrastructure to facilitate. What are we going to do, launch a container on every grades endpoint hit? Static analyze the grading code? No thanks.
The best sandbox is the client browser. Load up their browser with the raw data, run some code locally. This should be fast and performant.
Then maybe we also do nightly batch jobs for instructors (maybe in a lame container), or maybe we can get away with just shipping all the data to the instructor's browser as well (that would be sweet).
The text was updated successfully, but these errors were encountered:
Instructors want to grade in different ways.
One route is to say "No, this is not a feature/goal of DOJO. We will export to Canvas/other LMS, good luck!". But then you need a LMS, and historically we haven't reached for tools like Canvas because how hard is it to implement grading and deadlines? Probably this is not the path forward.
The issue of "arbitrary grading schemes" is that it's roughly equivalent to "arbitrary code". We could play sandbox games, and such, but ultimately when a student clicks on their grade page, they'd probably like to see the answer in <500ms. Sandbox games sound slow / a decent amount of code/infrastructure to facilitate. What are we going to do, launch a container on every grades endpoint hit? Static analyze the grading code? No thanks.
The best sandbox is the client browser. Load up their browser with the raw data, run some code locally. This should be fast and performant.
We could be sane and say "write javascript". But let's be real, we want to write python.
https://pyodide.org/en/stable/console.html
Then maybe we also do nightly batch jobs for instructors (maybe in a lame container), or maybe we can get away with just shipping all the data to the instructor's browser as well (that would be sweet).
The text was updated successfully, but these errors were encountered: