What actually happens to your files

When a file is saved, the app records the link between the thing on Canvas and the thing on your disk. That record lives inside your course folder itself, in a small hidden file called .canvas_sync.db. It is what makes a sync possible months later: without it, the app could only compare filenames, and a filename is a terrible way to identify a document.

Because the link is one-to-one, a file can never be quietly downloaded twice under two names. And because the record lives in the folder rather than in the app, you can move the whole folder to another drive, or another computer, and sync still works. Delete the folder and the app forgets it completely - there is no hidden copy of your data anywhere else.

The obvious place is the course's Files tab. The catch is that teachers can switch it off - and a lot of them do. On a course with the Files tab disabled, opening it yourself gives you a permission error, and a naive downloader would conclude the course is empty.

So the Files tab is only the first of several passes:

Even in All in One Folder a few things still get a folder of their own, because they'd be unusable otherwise: the contents of an unpacked archive (which have their own internal structure), lecture recordings if you asked for them separately, and Canvas content if you switch on the option below.

Keeping generated content separate

Things the app creates from Canvas - an assignment brief saved as a web page, a quiz, an announcement - can either sit inline next to the real course files, or be routed into their own tidy folders. Turn the separation on and you get:

Folder	What goes in it
Assignments/	The brief for each assignment, plus anything attached to it
Announcements/	Course announcements and their attachments
Quizzes/	Quiz questions and answers
Discussions/	Discussion threads
Syllabus/	The course syllabus
Submission Feedback/	Your grades, rubric scores and teacher comments
Pages/	Canvas pages
Links/	External links saved as shortcuts

An assignment brief isn't a document you can download - it's a web page inside Canvas. The app rebuilds these as self-contained HTML files you can open, search and keep after you lose access to the course.

Option	What you get	Attachments?
Assignments	The full brief and description, saved as HTML	Yes
Announcements	Each announcement as HTML, dated	Yes
Quizzes	Questions and answers as HTML	n/a
Discussions	The thread, including replies	n/a
Syllabus	The syllabus page as HTML	n/a
Submissions (Results)	Grades, rubric scores, teacher comments	Teacher's only

If your course uses Panopto, the app can find the recordings and pull them down in up to four forms. You choose any combination:

Output	What it is	Notes
Video	The full lecture as MP4	By far the largest option
Audio	Audio only, as MP3	A fraction of the size; enough to re-listen
Transcript	Plain-text transcript	Needs a model
Subtitles	Subtitles with timestamps	Needs a model

Recordings can either sit alongside the course material they belong to, or be collected in their own folder. You can also permanently ignore individual recordings - a guest lecture you'll never rewatch - and they stay out of every future sync until you restore them.

Conversions run after everything is downloaded, and only over the files this run actually fetched. A second run doesn't re-convert your whole folder.

Conversion	Applies to	You get	Original
Unpack Archives	.zip .tar .tar.gz	A folder with the contents	Replaced
PowerPoint → PDF	.ppt .pptx .pptm .pot .potx	A PDF of the deck	Replaced
Legacy Word → PDF	.doc .rtf .odt	A PDF	Replaced
Excel → PDF & data	.xlsx .xls .xlsm	A PDF, plus a _Data.txt with every cell value (.xls gets the PDF only)	Replaced
Canvas Pages → Plain Text	.html	A Markdown version beside it	Kept
Code & Data → .txt	50+ code and data formats - .py .js .sql .json .csv .yaml and more	A readable .txt with a header	Replaced
Gather Web Links	.url .webloc	One Compiled_External_Links.txt for the course	Replaced
Video → Audio	.mp4 .mov .mkv .avi .m4v	An MP3 of the audio	Replaced

Three details worth knowing

"Legacy Word" really does mean legacy

It converts .doc, .rtf and .odt - the formats that are awkward to open on a modern machine. A .docx is left completely alone, because it already opens everywhere. If you turn this on and your .docx files don't become PDFs, that's why.

Code files are renamed, not suffixed

analysis.py becomes analysis_py.txt - the dot turns into an underscore rather than the extension simply being appended. That keeps the original type visible in the name while guaranteeing the result can't collide with a real .txt that was already sitting there.

Office conversions need Office

PowerPoint, Word and Excel conversions drive the real desktop applications behind the scenes. If the relevant app isn't installed, those conversions are unavailable - the files download normally, they just aren't converted.

Turn on Unpack Archives and a zip is extracted. Turn on Code & Data → .txt as well, and you might reasonably expect the code inside that zip to be converted too. It isn't. Nothing inside an archive is ever converted - it's unpacked and then left exactly as it is.

This was tried the other way round first. Here is what happened on one real lecture zip:

Sync follows the identical rule, so a folder can never drift into a different state just because it was synced rather than downloaded.

Slides & PDFs only

Narrows the download to slide decks and PDFs, and nothing else: .pdf, .ppt, .pptx, .pptm, .pot and .potx. Datasets, images, videos and code are skipped - and so are Word documents. If your reading list is a .docx, this filter will not bring it down; use All files instead.

Maximum file size

Set a cap and anything larger is skipped. Useful on a metered connection, or when a course has a handful of enormous videos you don't want.

When you first download a course, your choices - subfolders or flat, which conversions, which Canvas content - are written into the folder alongside the file records. A sync reads them back and applies exactly the same rules.

That's why a synced file lands where you'd expect and gets converted the same way, months later, without you remembering what you picked. It's also why sync has no settings screen of its own: adding one would let your folder drift into a half-and-half state where some files followed one set of rules and some followed another.

The six categories

Every difference between Canvas and your folder lands in exactly one of these.

Before a sync touches anything, it compares three things: what's on Canvas now, what your folder's record says it saved, and what's actually on disk right now. Any disagreement becomes one of six categories - and three of them are unchecked by default, because they represent decisions you made.

Category	What it means	What happens	Default
New Files	On Canvas, not in your folder yet	Downloaded into the right place	Checked
Updates (Clean)	Canvas has a newer version and you haven't touched your copy	Replaced in place - same name, same folder	Checked
Edited locally	Canvas has a newer version and you've annotated or edited yours	New version saved as _NewVersion beside yours. Your file is untouched.	Unchecked
Deleted Locally	You deleted it; Canvas still has it	Nothing, unless you tick it. Quick Sync always skips these.	Unchecked
Deleted on Canvas	Teacher removed it; you still have it	Nothing at all. Your copy stays exactly where it is.	Info only
Ignored Files	You chose to permanently skip it	Never appears again until you restore it	Permanent skip

If sync only matched on filenames, renaming a lecture slide would make it vanish from the record and come straight back down as a "new" file - leaving you with two copies and a mess. So before comparing anything against Canvas, the app tries to recognise files you've moved or renamed:

1. Same name, new location. Dragged into a subfolder? Matched immediately.
2. Different name, identical contents. Matched on the file's contents and size, so a rename is recognised even if the new name looks nothing like the old one.
3. Similar name, close enough. A cautious last resort, used only when exactly one candidate fits.

When one of these matches, the record quietly updates to your new name and location, and the file stays up to date. No duplicate, no download.

You highlight a lecture PDF. You fill in the blanks on a worksheet. You add your own notes to a page. Then the teacher uploads a corrected version. Both of those are valuable, and the app refuses to choose between them:

This holds for converted files too, which is the harder case. If your annotated file is itself the output of a conversion, the protection applies to the file you edited, not the file that was downloaded - so re-running the conversion can't quietly regenerate over your work.