[MRG] hnn-core integration

[blakecaldwell]

This runs HNN simulations using the code in hnn-core. Significant changes to HNN parameter handling were required along new code to read and write data files coming from hnn_core objects.

Work remaining:

• Remove model visualization feature from HNN GUI
• Update installation docs to remove dependencies on pyopengl for model visualization
• Add hnn-core prerequisite to installation docs
• Dipole display after running multiple trials is not correct?
• Add CI testing base functionality
• Print status messages fromhnn-core to the simulation log window
• Correct the simulation shown in combobox after a run
• save_figs = 1 functionality
• Data can be held in memory after a simulation rather than writing to disk and re-reading.
• Open view windows (e.g. spike raster plots) without using POpen()
  • Will document the parameters that are used for each view window in docstrings
• Integrate somatic voltages/current calculation from HNN core

Closes #231

[blakecaldwell]

The PR is ready for testing! Optimization doesn't quite work as expected yet. The other known bug is with recording somatic voltages with cores > 1. I'll be sending out an email inviting others to test, but feel free to give it a go if you have everything installed already

[jasmainak]

@blakecaldwell what do you think of the following workflow?

• we tag the latest commit on master branch on github as v0.1 (or whatever is the appropriate version number). Maybe also create a branch called maint/v0.1
• we update the installation instructions to use the v0.1 branch/tag
• then we merge this branch into master

This way, you can have some peace of mind and we can iterate a bit rather than having to work off this huge branch. It's already looking like a great improvement to the code. We can iterate on the master branch then without breaking user workflows and when we make a new release we update the install instructions

[blakecaldwell]

@jasmainak yeah, I like that. As you say, we can iterate with new PR's and issues. They can be part of jonescompneurolab/hnn too

So the plan would be naming a new release v1.3.2. Since we've been in "production", releases have been > 1.0. I'll create a branch maint/pre-hnn-core to use in the docs. Also, for any potential compatibility issues that come up, this would draw the before/after line.

[blakecaldwell]

As discussed above, we are merging this to master to complete testing and debugging. Installation instructions for Mac and Windows on hnn.brown.edu will have people use release v1.3.2 which was created yesterday.

When we are happy with the post-integration code, we will create a new release and update the instructions.

The developers who are already running hnn from a git clone of master would only be affected if they ran a git pull. They probably will know what's going on if they do this and things look different.

[jasmainak]

There seems to be a duplicate set of installation instructions here: https://github.com/jonescompneurolab/hnn/tree/master/installer

Are these relevant any more? Because they are downloading the latest master

[blakecaldwell]

There seems to be a duplicate set of installation instructions here: https://github.com/jonescompneurolab/hnn/tree/master/installer

Are these relevant any more? Because they are downloading the latest master

Ah, yes. I've been keeping the new instructions in a separate branch (the link that I emailed out). Docs in master are the instructions that everyone sees at hnn.brown.edu.

Just created #280 with the new docs that everyone is using for testing.