It seems there are enough people out there still on Emacs 24 to warrant undoing
some of my recent moves away from eieio to cl-generic. For the near future I
will try to keep any changes that are incompatible with Emacs 24 to a
development branch. There is still probably plenty of opportunity to make the
code more future-proof for an eventual move away from eieio (I'm looking at you
oref and sref...)
Images weren't being save to json in the way they should have. Mostly had to do
with how the attirbutes were being ordered. This is still, unfortunately, a
nasty bit code but it still seems to work.
This reduces some of the complexity in testein.py, but unfortunately running
tests is still unreliable. Running batchwise tests don't work at all in Windows,
and running from inside emacs tests often need to be run multiple times before
they pass.
The worst offender is the delete notebook test, which will pass on usually only
1 out of 3 tries.
Testing seems to have revealed a couple bugs, so win??
EIN can communicate with the jupyter contents API synchronously or
asyncronously, with the default behavior the latter. For a time it has
been possible to force synchronous from the code, but now I have made
that a user-configurable parameter (see `ein:force-sync`).
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.
Fixing test suite for changes in IPython introduced in 3.x and Jupyter. Still a long ways to go, but some improvements in how we generate test notebooks.
Remove Islands Tests.ipynb, we didn't need it. Remove temporary
notebooks after running tests - note that this is difficult to test
under Windows, so I am guessing that this is the right thing to do.
Get ein:junk-new working again under IPython 2.x.
Some non-related work on fixing `ein:pytools-export-buffer`. Not quite
there yet, but I have an idea of what is going on.
Protocol for error reporting has changed in IPython 3.0 messaging and
ein was missing error messages.
EIN was also saving "*" for the notebook execution count and per
nbformat v4 execute_count can only be null or a number.
Work in progress.
Lots and lots of changes trying to clean up code so there are not so
many checks of IPython version. Also a significant amount of refactoring
all notebook server requests code to ein-contents-api.
Also moving away from notebook-id concept; instead use notebook path to
uniquely identify notebooks/content.
Made `ein:query-ipython-version` more efficient by caching results.
Start of a refactor of code for accessing Jupyter's new contents API.
First up an implementation of list contents.
New function for flattening lists borrowed from cl-alexandria (how did
we survive withoutthis in the before times?)
Updating author list here and there.
Calling ein:query-ipython-version with IPython 2.x running will result
in a request error since the api REST url is not implemented. Getting a
404 response is inevitable unless the IPython devs backport the api URL,
but I can make ein complain less when this happens.
IPython 2.x+ has a JSON response when creating a new notebook. We detect
this and correctly parse the info so we can open a buffer with the newly
created notebook.
Also more updating of test code.
ein:notebooklist-reload, which is called from success
callback of ein:notebooklist-delete-notebook, needs to
access ein:%notebooklist%. However it was set to nil
in ein:testing-delete-notebook-by-name just after the
delete button is "clicked". This was meant to detect
if the delete process is finished, but was not working.
To solve this problem, I add ein:notebooklist-after-open-hook
and use it to detect end of the delete process.
