Discussion: Possible features

[gordonwatts]

I made extensive use of this tool to help me make it through teaching ~400 students taking an intro physics/engineering course. It was a huge help - thank you so much for creating this and then putting in a place I could get at it so easily. For some context, I'm an experimental particle physicist at the University of Washington. Most of the questions I wrote were generated by python code and printed out. This made it very easy to create multiple versions of the question with different initial values, and possible to generate "reasoned" distractions.

I collected a few things that would have made my life easier as I was working. This issue is just a collection - if any of them seem like the right direction, a separate issue can be created to track their progress. Obviously, feel free to ignore any of these that are not interesting to the direction of the project!

• When duplicate answers are detected, print out all the possible answers. When the answers are generated by code it is often impossible to tell what lines are causing it. Seeing the answer in order, and then identifying the duplicates visual is better than removing the answers one-by-one to find the error.
• Pretty Print output of the questions (to a plain markdown file or html). If you have python code in you can't see the question until you've imported it into qti, which makes the round-trip time for editing a question much longer (see Is it possible to export the markdown to well-looked pdf or html? #35)
• Allow a set of common functions or constants defined in a python file to be imported into a question (see enable global functions for multiple questions #34 for a sample of what I'm talking about here)
• Allow image sizes to be changed. I did this with attributes, as you saw (see Allow image size to be changed #41).
• Title should be considered when deciding if a question is a duplicate. This helps when you are generating questions accross exams that are related, and so in the same file and identical other than the title label.
• option to not upload file. This may not be possible, but it seems images move with exam questions, not with their location. So perhaps the images can be encoded and not uploaded into the course bank. Repeated imports make the image directory a huge hairy unnavigable mess.

Again, thank you very much for this tool!

[gpoore]

Thanks for the suggestions!

The first two points seem to fall under general debugging. Perhaps there should be --verbose and/or --debug options to provide extra info, possibly also with more info saved by default.

I think your particular issue with import will be fixed once I fix handling of Python environments under Windows. I'm currently finishing adding support for LaTeX export of solutions, and then this will be next.

Can you say a little more about title in terms of finding duplicates? Are you using a single file as a way to generate a question bank rather than a single quiz?

Images: I don't know if there's a way to not include images, and then upload separately and have them detected correctly, and that's probably not something I'll have time to experiment with. With my personal workflow, I almost never look at the image directory and other things related to quiz details on Canvas, because I'm just treating it as an upload location for quiz content. However, it would probably be easy to add an option that converts all images into inline base64 encoded img. That could be a simple way to address this.

[gordonwatts]

The first two points seem to fall under general debugging. Perhaps there should be --verbose and/or --debug options to provide extra info, possibly also with more info saved by default.

This would be great... as long as one could eventually make sure just the markdown was printed out and not a lot of other verbose information. My idea here is to have this run in behind the user's back each time the user saves the file. You can then use something like vscode's preview mechanism to see pretty quickly what you wrote: it already has nice markdown rendering built in, etc.

I think your particular issue with import will be fixed once I fix handling of Python environments under Windows. I'm currently finishing adding support for LaTeX export of solutions, and then this will be next.

Ah, that sounds like a nice addition (the solutions)!. Ok - I'll check it out when it happens!

Can you say a little more about title in terms of finding duplicates? Are you using a single file as a way to generate a question bank rather than a single quiz?

Yes - we are using a single file to generate questions for a question bank. We need to name the questions differently in the title so that they can be identified in the question bank (Exam 1, Exam 2, etc.). And they are generated with common code, often (but not always).

Images: I don't know if there's a way to not include images, and then upload separately and have them detected correctly, and that's probably not something I'll have time to experiment with. With my personal workflow, I almost never look at the image directory and other things related to quiz details on Canvas, because I'm just treating it as an upload location for quiz content. However, it would probably be easy to add an option that converts all images into inline base64 encoded img. That could be a simple way to address this.

I would consider this a very low priority request. What generated this was some failed image loads during an exam, and the IT folks here telling us we were using the images incorrectly. We tried to explain that the images didn't need to be in any particular location, but we got nowhere. After that, we also noticed that our File directory was becoming quite full of duplicate images.

[gpoore]

There is now basic support for exporting solutions in Pandoc Markdown, PDF, and HTML formats in the dev version on GitHub (#35). See --solutions and --only-solutions. I may refine this a bit further before I release a new version and update PyPI, but I'm already using it myself. Soon I should be able to move on to some of the other issues.