| Date | 11 March 2020, 10:30 CEST |
The student programmers are doing a great job. Last week we achieved the coverage of all the steps of Tutorial 1. Some steps are checked in an abstract setting (e.g.: the co-simulation plot is not checked… We only check if the co-simulation stopped), but the progress looks good. We made a major leap in versioning updates already (from hundreds of deprecated to dozens). Nonetheless, we are currently dealing with more difficult updates. Last week we needed to debug functionality in the new-project.ts and found out the source maps for that and lots of other files were not being generated. This will keep us busy for a while. We have a new R&D project related to this. For now the focus will be on the cloud app.
There has been a burst of new activity on FMI3, which is making it difficult to track in the formal model. In particular, there are changes to the XML schema and changes to how variable aliases are declared which may produce fundamental changes in the model. I am therefore monitoring the Standard changes rather than actively trying to model them, while the standard is in flux.
Modifications are being made to a new “lspserver” branch of VDMJ to support the LSP protocol. I’m currently testing with the Eclipse lsp4e client, enabling one feature at a time to see what messages are sent - the Microsoft documentation for the protocol is not very clear (no examples), so we really have to progress by example. VDMJ itself is currently proving relatively easy to change to add this support, but we’re starting with the simplest features…
I spent my time on two things.
(1) Reading the paper on the harvest coach recommended to me by Hugo as a source of simple examples. This is a very nicely written paper. The examples for the tolerance language are indeed very simple but also very helpful.
(2) I read a series of papers in the list sent to me by Claudio. I found one particularly useful: “Specification-Based Monitoring of Cyber-Physical Systems: A Survey on Theory, Tools and Applications”. This led me to some papers in systems biology that have a version of linear temporal logic with a satisfaction relation with degrees of violation.
I found that the ideas in (2) give an elegant formulation of the tolerance constraints in (1). What seems particularly good is making tolerance a first-class citizen: “the combine violates the position validity requirement by no more than 4 m”.
25 March 2020, 1030 CET