Page Comparison

General updates

• MT – I’ll be offline afternoons Eastern time next week.

• JW – Thanks for continuing to do my job - TK releases, PR porting, and user support - the last few weeks.

• JW – Now that vsites are merged, I'm working on topology/hierarchy interop. Tests are passing now but I still need to trawl around and figure out what details are left to do.

  • RGowers is starting to work on improving protein loading speed. May open a PR to smarts_to_netwrokxnetworkx_rdkit branch

• JW – Anonymous tip that industry may want a bespokefit single file installer

• JW – OpenFF units prints out OFFXMLs without * signs, and with Angstrom symbol - Not reverse compatible. Can we fix before alpha?

  • Difference is at (0.10.5 code)

    Github link macro
    link https://github.com/openforcefield/openff-toolkit/blob/master/openff/toolkit/utils/utils.py#L207-L233

    turning to (topology-biopolymer-refactor)

    Github link macro
    link https://github.com/openforcefield/openff-toolkit/blob/topology-biopolymer-refactor/openff/toolkit/utils/utils.py#L186-L187

• Still using supercomputer at UCI?

  • MT posted on #qcfractal-compute, giving TG just about everything back

PR and issue clearance

• EP 0005b

  • (JW – Removed periodic coulomb and closed)

• Working on updating electrostatics to be compatible with 0.4 spec

  • MT – I’ll push on getting the relevant changes into the toolkit and interchange.

  • MT – I don’t like how versions are stored as floats. Could open a PR to improve this

  • JW – It’d be fine to store these as version objects. I’d love a PR that does that.

• ChargeIncrementHandler bugs - Lots of weird cases that are only tested in Toolkit unit tests, showing up as failing when using interchange in create_openmm_system. Fixing these in Interchange - MT is working on this now.

• Lots of other little things uncovered, not a big deal.

• JW – Ok to turn off LGTM on toolkit?

  • MT – Yes please. And I think we should turn it off on all repos

  • (Turned off LGTM, verified that MT also has permissions to enable/disable)

• create_openmm_system can take a toolkit registry, from_smirnoff can't

  • Door #1 is to not accept toolkit_registry

  • Door #2 is to copy toolkit behavior

  • Door #3 is a big refactor to do like context management

• JW – Let’s handle this in create_openmm_system, before we call from_smirnoff - Can use context manager trick from

  Github link macro
  link https://github.com/openforcefield/openff-fragmenter/blob/4a290b866a8ed43eabcbd3231c62b01f0c6d7df6/openff/fragmenter/utils.py#L97-L123

  • MT – If you like, I could open a PR for this.

  • JW – That would be great - It’ll realistically actually get done if you work on it.

  • MT – WIll do, I’ll include this in the PR to remove from_interchange