Nico Hoffmann ෴.
Nico Hoffmann ෴.
@afbora We cannot delete the tmp file on cancel. But we clean up the files with a garbage collection.
@lukasbestle makes sense. And as written in the initial post, I think we should have a few review rounds as well. Now getting the concept right, later adding unit tests...
@afbora That's what we are doing here. Just not the system temp folder as that doesn't persist across requests. So the first parts aren't there anymore when the last chunk...
@afbora the PR currently is not functional as it's in the middle of implementing changes based on Lukas' feedback and some troubles I encountered (see our comments discussion). Sorry that...
@lukasbestle @tobimori are you thinking of something different than the existing accept maxsize option?
@lukasbestle added some more validations – skipping MIME types (will still checked for the full file), but at least basic ones as allowed extensions.
@lukasbestle I am not sure we can follow your suggestions for `maxsize`. This would require us parsing the blueprints on each of those requests (Panel view, chunk upload). And given...
@lukasbestle I added a check on the backend for `maxsize`: both, checking the total length from the header as well as existing tmp plus chunk size.
Would be great indeed. I've rebased and squashed the commits already.
@afbora good catch. Can reproduce myself. Somehow already after the first chunk they get moved to the content folder. But need a fresher mind to investigate why.