Zenodo chat

BW: can immediately get Zenodo DOI with empty upload

Process

• Create an empty Zenodo record

• Create the dataset with the DOI

• Upload the dataset

• Redownload the results and upload to Zenodo

Can also modify the description later on MolSSI

• Create the dataset and upload

• Redownload, modify description, and upload onto Zenodo

• (still runs into chicken and egg problem?)

YDS automated upload

• BW: doesn’t obtain DOI until publication

• BW: You may need to serialize the records from MolSSI

• LW: The MolSSI side won’t provide all properties, e.g., wave function, unless requested

• LW: Spice lives online as hdf5 file that does contain everything

• Jen will ask Ben about using hdf5

• BW: can prereserve_doi: bool in the Zenodo API

QCA-DS FF Releases:

• LW: table at the top of the README / before all the others sounds good

• Discussion of FF Release table:

  • Could do one table with a row for each force field,

    • one with the FF name (and link to fitting repo, e.g. )

    • Opt dataset

    • TD dataset

    • Elements

    • Zenodo

      • Upload all training data into 1 record, tidier. Makes CI harder

      • Discussion – could we do multiple directories in a PR?

      • JCl: can try to do Sage 2.1 in 1 PR

  • (Discussion about openforcefield.org landing pages – the 2.0.0 page is more of a news post, other FFs not guaranteed to have one)