Grading source code in an online environment

It has never occurred to me to use printouts of code, even though I have an older colleague who does this. To me, it seems a little odd as an approach and inefficient.

That said, if you wish to have a similar workflow digitally, you might want to consider using a tablet computer with a pen, Microsoft OneNote, and then saving the final output as a PDF. This will allow you to draw, scribble, and write freely on their submitted code. As a side benefit, you'll also be able to put in links, pictures, or other diagrams, assorted and sundry as you find useful.

Typically, I provide feedback either in bullet points or side-by-side in person ("in person" in this context means using screen share in Zoom).

If you can create PDF files (as you indicated in comments) you can use Adobe Acrobat Reader (the free one) to add comments from a keyboard into the code and save it. This is what I use for most student work, listings, dissertations and so on. You can save it and re-upload to the VLE or whatever feedback mechanism being used.

Of course, if you have a tablet device or even a sketching pad attached to a desktop you can add more freehand annotations in the old style.

However, I have moved from annotating code for a more test-based development mechanism for giving coding feedback. The students now have to run the test scripts I write and upload the results rather than the code at intermediate stages for formative feedback. I only feedback on the code at the summative stage. I find that, with >250 students, I need to let the computer do more of the work whilst still trying to give some form of helpful information to the students.

There are, of course, pricier tools specifically designed for scribbling that others may mention.

Another way of doing things is to interact with the students' actual code base and not a listing, using something like Visual Studio Live Code Share. This give feedback at the actual point where they start to deviate from common sense. Some colleagues are doing this technique.

You can try PDF Viewer Pro by PSPDFKit, saw it in the top list on iOS App Store a few days back, looks like it does what you need. I say iPad because I think it allows you easier to create circles and lines than using a mouse on a notebook.

Note this answer does assume a batch scanner is available.

If you insist on working with printed pages, you could prevent doing (much) manual work returning the files if you either:

1. Keep track of the order of the printed files, correct them, put them in a batch scanner and automatically return the corrected files. Though I could imagine this is not an option due to risk of mixing files due to scanner/printing glitches.
2. Another option could be to print a qr code on the page,print the pages, correct the pages put the corrected papers back into a batch scanner, and process the images using a qr decoder to allocate the files accordingly, this way you are (almost, depending on how good the qr scanner is) certain that the correct corrected papers are returned to the right students. You could also get a notification on the papers that were not scanned correctly since you know how many you printed and hence should scan.
3. If you don't want to be searching for a place to put the qr code in their paper, you could give the students a latex template, or convert their code into a latex template with a qr code belonging to their study number.

In my experience pyzbar can be a bit tedious to set up but I found it to practically work identifying qr codes in printed- and scanned pages. Not claiming this is the best solution, I think there would be easier solutions and like DarcyThomas answer better.