2020 12 07 Open NEST Developer Video Conference

previous | list | next

Agenda

1. Welcome
2. Review of NEST User Mailing List
3. Project team round
4. In-depth discussion
   • Use squash and merge for PRs? #1856
   • developing connection syntax

Mailing list

• https://nest-simulator.org/mailinglist

Project team round

Here we discuss topics that need broader attention, for example questions that came up but are outside a single project's scope, larger planned changes/PRs that affect all teams or pending work that is blocked by external factors.

1. Models / NESTML
   • minor updates
   • many reviews pending, e.g. Siegert neuron
2. PyNEST
   • no major updates, developing connection syntax (in depth)
3. Kernel *
4. Installation
   • nothing major to report
5. Infrastructure
   • mirgrating from Travis to GitHub Actions
   • still waiting for response from Travis-team
6. Documentation
   • NEST-Server documentation is being updated, pr is being prepared
   • [#1843](Doc build system), looking into issues with undefined variable on RTD
7. EBRAINS
   • installing NEST on Google-Collab, any experience?
8. (Feature) Random number generation
   • writing documentation
   • benchmarks being done at the moment, investigating recent drop in performance
9. (Feature) Automated Testing
   • behavior of make installcheck needs to be explained in the docs. HPC frontend nodes may have no MPI installed, while compute nodes are perfectly fine.
   • will change to usual behavior of skipping checks, if runtime environment is not set up correctly. (as for libneurosim, libcsa, libmpi...), continue discussion in issue
10. (Feature) Extension Module System
    • discussion about separating each model in separate modules (NESTML generated modules would favor having compatible synapse and neuron models in the same module)
    • implement #1717 then do fine-grained modules if/when necessary.

In-depth discussion

Use squash and merge for PRs? #1856

• for small PRs, fixing syntax, etc. becomes much easier to track in history
• for medium to large PRs this prevents bisecting history afterwards
• more difficult to track contributions by different authors in case there are different manual co-authored by: entries
• hierarchy between author and co-author is not necessary.
• difficult to separate small(squash) and medium(non-squash)
• Github already treats source-branch squashed force-pushes correctly, i.e. developers can squash commits even after filing the PR.

=> Only squash if requested by the author and squash makes sense. Generally stick with plain merges to preserve smaller changes for bisections.

Naming of main branch of the repository

• requires testing
• propose date
• switch at coordinated point in time
• https://github.com/github/renaming

developing connection syntax

• working on updated pynest connect syntax hakonsbm/nest-simulator:connection_semantics_prototype
• handle creation and connection generation on C++ level
• hand Projection objects to C++ from Python level
• Models handled as classes on Python level and instantiation is possible to parallelize on kernel level.

=> Separate break-out VC next week