Brandon Bocklund

I re-ran the ternary example, it seems fixed and significantly faster. In fact, I was getting an error on develop, some Gibbs phase rule violation in array sizes, I believe....

@wahab2604 are you still having issues with plotting phase diagrams with phases that contain associates?

Great, thanks for the update! In the current implementation of `binplot` your ability to resolve small features in composition (and to some extent, your ability to resolve miscibility gaps) is...

It turns out that a point density of 10 million was enough to get a miscibility gat at 2105 K, but it did take almost 20 minutes for a point...

One way you can work around it is by providing a custom set of points for the LIQUID phase that adds some carefully chosen points that trigger new composition sets...

@richardotis how was this confirmed? Did you have the database or look at another system. I just checked the script above against `develop` and the results look pretty good using...

This is related to #419, with the core issue being that species names are basically cosmetic in DAT files

Option 2 sounds fine to me. Right we haven't been building tagged versions for awhile anyway. We could set up another action that is triggered on tags to build, commit,...

Can be triggered by pushing an empty commit to the website branch: ```bash git commit --allow-empty -m "Bump website to trigger deployment" ```

I think it’s caused by using the Phase constituents unfiltered for the active components [here](https://github.com/pycalphad/pycalphad/blob/373d1212c2a350239195ea4c7b984d26648fec47/pycalphad/model.py#L1023) IMO, both cases should pass if `'D'` is inactive, but fail if `'D'` is active.