CI failures with tornado 6.0a1

[takluyver]

Tornado 6.0a1 was just released, and is being picked up by Travis CI because we use the --pre option to install dependencies. This causes a couple of test failures.

We could drop the --pre flag and continue testing with Tornado 5.x for now. But it's useful to pick up problems when dependencies are in beta or RC.

Tracebacks:

Traceback (most recent call last):
  File "/home/travis/virtualenv/python3.6.3/lib/python3.6/site-packages/tornado/web.py", line 1697, in _execute
    result = method(*self.path_args, **self.path_kwargs)
  File "/home/travis/virtualenv/python3.6.3/lib/python3.6/site-packages/tornado/web.py", line 3174, in wrapper
    return method(self, *args, **kwargs)
  File "/home/travis/build/jupyter/notebook/notebook/nbconvert/handlers.py", line 91, in get
    exporter = get_exporter(format, config=self.config, log=self.log)
  File "/home/travis/build/jupyter/notebook/notebook/nbconvert/handlers.py", line 61, in get_exporter
    from nbconvert.exporters.base import get_exporter
  File "/home/travis/virtualenv/python3.6.3/lib/python3.6/site-packages/nbconvert/__init__.py", line 7, in <module>
    from . import postprocessors
  File "/home/travis/virtualenv/python3.6.3/lib/python3.6/site-packages/nbconvert/postprocessors/__init__.py", line 5, in <module>
    from .serve import ServePostProcessor
  File "/home/travis/virtualenv/python3.6.3/lib/python3.6/site-packages/nbconvert/postprocessors/serve.py", line 19, in <module>
    class ProxyHandler(web.RequestHandler):
  File "/home/travis/virtualenv/python3.6.3/lib/python3.6/site-packages/nbconvert/postprocessors/serve.py", line 21, in ProxyHandler
    @web.asynchronous
AttributeError: module 'tornado.web' has no attribute 'asynchronous'

...

Traceback (most recent call last):
  File "/home/travis/build/jupyter/notebook/notebook/services/kernels/tests/test_kernels_api.py", line 176, in test_connections
    ws = self.kern_api.websocket(kid)
  File "/home/travis/build/jupyter/notebook/notebook/services/kernels/tests/test_kernels_api.py", line 65, in websocket
    return loop.run_sync(lambda : f)
  File "/home/travis/virtualenv/python3.6.3/lib/python3.6/site-packages/tornado/ioloop.py", line 531, in run_sync
    return future_cell[0].result()
tornado.websocket.WebSocketError: Non-websocket response

[takluyver]

I opened an issue on pip to request the ability to install betas but not alphas: pypa/pip#6110

[takluyver]

The first traceback there is a failure importing nbconvert; I've made PR jupyter/nbconvert#937 to fix that. No idea about the "Non-websocket response" at the moment.

It's probably sensible to constrain tornado<6 for a while, as I've done in #4310.

[liavkoren]

I just ran into this issue, should I install of master or do you guys expect to cut a release soon?
(Also, thank you for the time and attention you guys put into this package. : )

[cguethle]

tornado 6.0 dropped today and broke our stuff with this issue. Seems this was not resolved prior to release.

[danking]

Also appears to be affecting us.

[danking]

@takluyver Looks like jupyter just needs to publish to PyPI. #4310 wasn't included in the last publish on December 17, 2018.

[spex66]

Since it is atm only in master, here is how I limited tornado to <6 until the fix is deployed to conda packages.

Warning: I'm a conda noob and googled this stuff, but it worked for me.
(i.e. I still don't know where conda saved this environment change to)

I created one yml file:

name: base
dependencies:
   - tornado<6

And updated conda base (aka root) environment with it like so:

conda env update -f 190304-root-env-patch-tornado.yml

thx to SO post:

• https://stackoverflow.com/a/32799944/1104502

Now I need to remember, when this is fixed in conda and to remove this patch again :)

[Tantael]

That why you should never use undefined package version on your production :D

[amadeann]

The solution from @spex66 worked from me as well. Thanks!

[edgarriba]

same error here 😭

[lkhphuc]

I was not able to run the newest version install with conda install jupyterlab.
After that I downgrade tornado using conda install tornado=5.1.1 and it was good to go.