Almost a direct copy and paste from ein-ac.el. Will have to
iterate on this more than a few times to figure out a good system
for generation completion candidates for company mode.
No longer set buffer-file-name for ein:notebook buffers. This was
causing unexpected and potentially disastrous (as in data corruption)
behavior.
Some updates to travis configuration and testing. Still a long ways
from having that working, though.
Newly created notebooks no longer error on saving - bug in how
kernelspec metadata was being filled in.
Also a number of miscellaneous fixes to indentation that have
inexplicably appeared in the code.
Some small fixes for typos and catching up on version info.
Only six months late. Jupyter changed callback signature for
complete_request, breaking ein's autocomplete code. Missed this for so
long because I am lazy and almost exclusively used jedi for
autocompletion. Ugh.
Code wasn't calling `ein:notebook-save-notebook-error` when content
saving failed. Small detail, but would result in "Notebook saving"
eternally hanging out in notebook message display.
Notebooks created directly from ein store image data differently than
notebooks created from the web interface. We now try to fix that when
saving notebooks.
Somewhat hackish solution by defadvicing json.el code to
overcome possible bug in said code. Apparently
`json-encode` does not handle nested empty hash tables properly.
Be more forgiving in detecting version; this change is mostly so EIN
works with recently released Jupyter/IPython 4.0.
Adding some additional comments to documentation to note that EIN seems
to work with Jupyter.
New commend `ein:notebook-reconnect-kernel` which will close then reopen
the websocket connection to a running kernel - without killing the
kernel.
Was not correctly killing and restarting the kernel in
`ein:kernel-restart.`
Fixed `ein-restart-kernel` to use REST API to kill kernel, then
restart/reconnect, again using the REST API.
Got rid of the `ein:kernel-stop` function. Seems confusing to have
'stop' and 'kill' methods running around.
Support recently added feature to websocket for including cookies in
connection header. This allows us to connect to secured IPython 3.x (and
2.x I hope) servers.
This should fix an obscure (well, at least no one has created an issue
for this yet) failure trying to open a notebook path that contains
characters like (, or ), etc. that need to be escaped in the REST API
call.
Stopping a kernel now doesn't automatically close a buffer, nor does
closing a buffer automatically stop a running kernel. This is
functionality more in line of what EIN supported for IPython 0.x and
1.x.