2017 11 13 Open NEST Developer Video Conference

previous | list | next

Agenda

1. Welcome

2. NESTML Talk: Construction of a DSL with Python

   A framework for the automatic model code generation from a domain-specific language (DSL) like NESTML was presented. The complex process of model implementation is broken down to separate automatized steps which are small and clear to understand, develop and maintain. The automatic translation from a DSL to model implementations has the potential to make the code more flexible and less prone to user implementation errors.

   The possibilities of verification of the generated code and the degree of flexibility of a DSL were discussed. Testing and verification are however a more general problem and should probably not be handled from within the DSL. On the other hand, a more integrated solution would enhance the users perception of neuroscience models and accompanying validation/tests.

3. Towards NEST 3: Updates from Ås

   • first commits for NEST3
     • current failing build probably due to travis log length limitation
     • log files of building external tools could be stripped
   • GIDCollections
     • have shown to be very powerful, many things become very compact
     • capabilities extended for better support for topology
   • topology
     • now mosly working with subnet-free NEST, almost complete
     • so far it's possible to have many nodes in a given location within a layer. This feature will likely be dropped. Nodes with same location can still be implemented by placing them in separate layers with same position.
   • objectification of PyNEST
     • GitCollectionDatum objects on PyNEST level
       • subclass for topology layers will be added
     • will enhance usability

4. Review of NEST User Mailing List

   • no unhandled mails.

5. Review of open Github Pull Request

   • PR #853
     • versions would have to match exactly since there is no defined api on sli or c++ levels yet
     • maintenance and keeping two separate packages in sync would probably be problematic
     • packaging NEST for further distributions would be a prerequisite. Installing nest from pip is likely to be problematic in many setups.

6. Review of open Github Issues

   • skipped due to lack of time