[MRG] Add nbresuse==0.3.3 (full freeze.py)

[manics]

Adds nbresuse==0.3.3, followed by running freeze.py

JupyterLab automatically displays the metrics provided by nbresuse, it doesn't need a separate extension. Note that as mentioned in jupyterhub/binderhub#1097 (comment) there are alternative JupyterLab front-ends for nbresuse, for instance https://github.com/jtpio/jupyterlab-system-monitor

Closes jupyterhub/binderhub#1097

Screenshots from testing locally with docker run -e MEM_LIMIT=200000000 ...

Notebook:

Lab:

[betatim]

I think this will be a useful addition. Going to wait and see if someone else has an opinion.

---

On a meta level it makes me wonder if we should start thinking about "repo2docker" and "repo2docker binderhub style". The latter would be r2d + all the custom extensions and packages we are adding that are useful when running in a BinderHub. Or maybe it isn't needed as most people use repo2docker via a BinderHub any way.

[manics]

It looks like Travis is failing on Python 2.7:

+ /srv/conda/envs/kernel/bin/ipython kernel install --prefix /srv/conda/envs/notebook
Traceback (most recent call last):
  File "/srv/conda/envs/kernel/bin/ipython", line 7, in <module>
    from IPython import start_ipython
  File "/srv/conda/envs/kernel/lib/python2.7/site-packages/IPython/__init__.py", line 49, in <module>
    from .terminal.embed import embed
  File "/srv/conda/envs/kernel/lib/python2.7/site-packages/IPython/terminal/embed.py", line 18, in <module>
    from IPython.terminal.interactiveshell import TerminalInteractiveShell
  File "/srv/conda/envs/kernel/lib/python2.7/site-packages/IPython/terminal/interactiveshell.py", line 16, in <module>
    from prompt_toolkit.document import Document
  File "/srv/conda/envs/kernel/lib/python2.7/site-packages/prompt_toolkit/__init__.py", line 16, in <module>
    from .interface import CommandLineInterface
  File "/srv/conda/envs/kernel/lib/python2.7/site-packages/prompt_toolkit/interface.py", line 19, in <module>
    from .application import Application, AbortAction
  File "/srv/conda/envs/kernel/lib/python2.7/site-packages/prompt_toolkit/application.py", line 3, in <module>
    from .buffer import Buffer, AcceptAction
  File "/srv/conda/envs/kernel/lib/python2.7/site-packages/prompt_toolkit/buffer.py", line 7, in <module>
    from .auto_suggest import AutoSuggest
  File "/srv/conda/envs/kernel/lib/python2.7/site-packages/prompt_toolkit/auto_suggest.py", line 13, in <module>
    from .filters import to_cli_filter
  File "/srv/conda/envs/kernel/lib/python2.7/site-packages/prompt_toolkit/filters/__init__.py", line 33, in <module>
    from .base import *
  File "/srv/conda/envs/kernel/lib/python2.7/site-packages/prompt_toolkit/filters/base.py", line 5, in <module>
    from prompt_toolkit.utils import test_callable_args
  File "/srv/conda/envs/kernel/lib/python2.7/site-packages/prompt_toolkit/utils.py", line 9, in <module>
    from wcwidth import wcwidth
  File "/srv/conda/envs/kernel/lib/python2.7/site-packages/wcwidth/__init__.py", line 12, in <module>
    from .wcwidth import ZERO_WIDTH  # noqa
  File "/srv/conda/envs/kernel/lib/python2.7/site-packages/wcwidth/wcwidth.py", line 82, in <module>
    from backports.functools_lru_cache import lru_cache
ImportError: No module named functools_lru_cache
Removing intermediate container e2a8fd2ebbd3
The command '/bin/sh -c bash /tmp/install-miniforge.bash && rm /tmp/install-miniforge.bash /tmp/environment.yml' returned a non-zero code: 1

which may be due to the change to wcwidth:

-  - wcwidth=0.1.8=py_0
+  - wcwidth=0.2.3=pyh9f0ad1d_0

Upstream diff: jquast/wcwidth@0.1.8...0.2.3
How should I deal with this?

[manics]

For comparison I've also got a separate branch with a minimal commit that only adds nbresuse and psutil, this passes Travis: manics@f53c291

[betatim]

Seems like a bug in wcwidth no? Can we pin to a lower version that still works?

[manics]

I think it's a bug in the conda-forge wcwidth package (upstream added a new dependency for python<3.2 which is missing from the conda package), I've pinned it for Python 2.7 only.

The remaining CI failure looks unrelated.

[betatim]

Restarted the CI. It seems sometimes the "does the memory limit work" tests fail. It seems like some of the nodes our tests run on support it and some don't :-/

[bitnik]

@manics thanks for adding this! may I ask why you didnt add v0.3.4 instead? I mead, did you simply miss it or is there a specific reason? I am just curious :)

[bitnik]

okay, v0.3.4 has a problem with jupyterlab: jupyter-server/jupyter-resource-usage#36