dcc.Upload errors with big files

Context

• Creating an app where end users can upload files >1GB with dcc.Upload.
• The app code is the same as the first sample app of the doc page: https://dash.plotly.com/dash-core-components/upload
• max_size is set to -1

Issues

• If I run the app with gunicorn or in my local env, some time after I upload the file the tab freezes or I get an internal server error (NOT a timeout because dcc.Upload has some kind of async functionality). > This is similar to what this issue describes: https://github.com/plotly/dash/issues/1728

- If I run the app with Python in a Dash Enterprise workspace, I get a callback error which is kind of weird. If you look at the callback code and the print statements in the terminal, we only get two values despite having three Inputs. The first one is probably an empty string or list, which causes the if statement to “fail” (since list_of_contents is not None), so the error we see is related to code after the if statement. The real problem is that, even if the filenames and dates get passed correctly, the content is empty > This issue has been described before in the forum: (https://community.plotly.com/t/upload-500mb-csv-but-content-is-empty/83297.

Additional context

• I've tried dash-uploader (link) and it works okay but I’m concerned about (1) privacy (all files are accessible by all users by default, as opposed to session-specific storage) and (2) memory (how does saving many big files affect the app memory/performance)
• dash-uploader is not an official component and it’s not maintained, which raises some concerns regarding production environment.

Request

• dcc.Upload being able to handle big files (500MB, >1GB, etc).
• If the file is not been processed correctly (list_of_contents is empty) show a more helpful error.

Start with the dcc.Upload documentation sample and reproduce the behavior using files over 500MB in the local, gunicorn, and Dash Enterprise environments described. Compare the callback inputs when content is empty or missing. Done means large uploads are handled reliably and an unsuccessful upload produces a helpful error.