Conclusion & Questions

Thanks for listening.

• The core functionality of ASR has been reimplemented.

• Dependencies are no longer given explicitly.

• requires / creates keywords have been removed.

• All recipes must take whatever relevant arguments they need. This is important for the caching mechanism, but also for general usability.

• A new concept of SideEffects have been introduced.

• The asr.database functionality will no longer be recipes. They will have their own CLI interface, ie., asr database fromtree ..

• These new changes have been implemented in asr.gs, asr.magstate, asr.magnetic_anisotropy, asr.relax, asr.structureinfo, asr.stiffness.

These are some questions you can think of

• There is now a greater disconnect between the users and the data. Before it was directly visible through the files in the folder, but now they are hidden.

• Do we want do delete all unregistered side-effects?

• Defect guys: We need to be able to migrate your data as well. Is there anything special about your data compared to the C2DB data?

• Multilayer guys: Same question to you.

• As you might have noticed I changed the @ to :: ie results-asr.gs@calculate.json. It is easier (for me) to type :: but we can skip this change if you’d like.

An outlook

• Me and JJ are thinking about how workflows are supposed to work in this new framework. The problem is that we used to depend on the existence of resultfiles. But they no longer exist.

• The calculator interface is not solved yet. Ask is currently thinking about/working on this.