Starting 0.2.0 rc.1 (Supersedes #29)

[pc494]

---

name: Starting 0.2.0 rc.1 (Supersedes #29)
about: As previous, with details to follow.

---

Release Notes

major
improvement & bugfix & developer change

Work in progress?
(more to be added)

• Kirkland Scattering (move to an issue, and do in a separate PR)
• PR template
• debeye waller docstring is slightly wrong (should be dict {str:float} or float)

[dnjohnstone]

I think we need to address #11 in this one to sort saving/loading of libraries etc.

Probably using an hdf5 format is the best route - @stefsmeets I know we discussed this in the past, do you have any opinions / suggestions for this?

[stefsmeets]

Probably using an hdf5 format is the best route - @stefsmeets I know we discussed this in the past, do you have any opinions / suggestions for this?

I agree hdf5 would be most flexible. I tried pickling (also using dill), but it crashed out on me. But I'm worried that hdf5 will be difficult to maintain, because a lot of custom attributes are being stored on different objects (also on pyxem). Hyperspy's BaseSignal already has a .save method, can we not make use of that somehow?

[dnjohnstone]

Yeah, in pyxem we intend to mostly use the hyperspy.save / load method that already exists.

In the case of simulated libraries though, these things are generated by diffsims, which we're trying to keep a lighter dependency list for. i.e. it does not depend on hyperspy and so we're going to have to do something different...

[dnjohnstone]

Is this PR dead @pc494 ?

[pc494]

Yep, I'll pull out the useful stuff somewhere else.