Anthony Gégo

lighttpd is known to have behaviour changes even between minor release, but it also can be due to the migration to flask. It would be worth looking at the logs...

Indeed, but the URL is generated from the FastCGI parameters that are passed to the app by the webserver. The Flask ``url_root`` is computed from the ``SCRIPT_NAME`` (tweakable) and ``PATH_INFO``...

Given your logs it's highly probable that lighttpd sets ``SCRIPT_NAME`` to ``/inginious-webapp`` by default, which is the one specified in the lighttpd config file (`fastcgi.server = ( "/inginious-webapp" =>`). You...

The complete information are historically fetched from the container and stored in database only for administrators. This is done here : https://github.com/UCL-INGI/INGInious/blob/master/inginious/frontend/pages/tasks.py#L210 This was done to avoid wasting valuable database...

Hello, Indeed, the length should be between 1 and 30 *chars*. The ``openssl rand -hex 16`` command outputs 16 *bytes* in hexadecimal format: that means 32 chars. It's too long...

Please separate a) and b) so that the review of b) can be done easier, as a) review should be quite straightforward. Please also fix the remaining conflict.

Can you rebase this PR now ?

As already mentioned, I think the ``inginious-webapp`` script should take an additional argument to make the protocol explicit.

Interesting suggestion. Actually there was no prior thinking about such a case. Changing the hash method will require an automatic migration mechanism for older accounts and ideally make this completely...

Useful yes, but probably not as simple as it may seem. There is no modeling layer in INGInious for now and many checks are done in the app logic. Writing...