During install of Mediagoblin following stable media goblin release 0.90 the following error is received: (Instructions to use) ​http://mediagoblin.readthedocs.io/en/stable/siteadmin/deploying.html OS: Debian 8 / Devuan Jessie

Running Sphinx-1.6.5/setup.py -q bdist_egg --dist-dir /tmp/easy_install-NMpkET/Sphinx-1.6.5/egg-dist-tmp-bQS0kd error: Setup script exited with error in Sphinx setup command: Invalid environm$ python_version<"3.5" Makefile:123: recipe for target 'bin/python' failed make: * [bin/python] Error 1

I was able to work around this by ./bin/easy_install sphinx==1.6.1

1.6.1 seems to be the earliest sphinx to not have this error.

Note there is another issue here.

the first time you receive the error it is when you use this command: ./bootstrap.sh && ./configure && make

however the second time you run that command, the error does not appear. it only appears when you run the update/upgrade command listed on the deployment page. that command is: git submodule update && ./bin/python setup.py develop --upgrade && ./bin gmg update

I would expect the error to appear under the bootstrap and configure commands.

I am currently hosting my GMG instance behind a nginx proxy using the following line:

proxy_pass http://localhost:6543;

Since the fastcgi configuration is totally not working (flup errors) this is currently the only way I have to host my GMG. It is behind a secured nginx server with letsencrypt support. However API URLs are returning ​http://localhost:6543/api/... urls because GMG does not know it is behind a proxy.

I managed to make everything work with the following nginx config:

location /
{
    proxy_pass http://127.0.0.1:6543;
    proxy_set_header Host       $host;
}

And this modification in app.py in order to enforce https:

request.environ['wsgi.url_scheme'] = 'https'

However this is hacky. I am not familiar enough with wsgi stuff to propose a patch but hosting via proxy should be way simpler than this hacking. Maybe propose new configuration parameters such as ENFORCE_HTTPS or stuff.

The celeryd executable is no longer available, and thus calling ./lazycelery.sh will fail.

See attached file for proposed solution. User needs to start an AMQP server, such as RabbitMQ, in order for the broken to start.