Issue with NF and functions with arrays of records

[casella]

Please consider the attached MWE. Try to simulate the model Elements with the NF. The following error is reported:

[1] 16:42:16 Translation Error
[C:/dev/OM64bit/OMCompiler/Compiler/NFFrontEnd/NFEvalFunction.mo: 389:9-390:53]:
Internal error NFEvalFunction.applyReplacementCref
could not find replacement for mats[i].x

[Karim.Abdelhak]

I cannot reproduce the problem simulating attached .mos script. It runs fine under the current nightly, am i missing something?

true
""
true
""
true
""
record SimulationResult
    resultFile = "/home/kab/Repo/FH/promotion/modelicaLibs/NF/TestArrayRecords.Element_res.mat",
    simulationOptions = "startTime = 0.0, stopTime = 1.0, numberOfIntervals = 500, tolerance = 1e-06, method = 'dassl', fileNamePrefix = 'TestArrayRecords.Element', options = '', outputFormat = 'mat', variableFilter = '.*', cflags = '', simflags = ''",
    messages = "LOG_SUCCESS       | info    | The initialization finished successfully without homotopy method.
LOG_SUCCESS       | info    | The simulation finished successfully.
",
    timeFrontend = 0.06924037000000001,
    timeBackend = 0.005097715,
    timeSimCode = 0.001241987,
    timeTemplates = 0.002963645,
    timeCompile = 0.473782734,
    timeSimulation = 0.012207977,
    timeTotal = 0.564635768
end SimulationResult;
""

[perost]

Replying to Karim.Abdelhak:

I cannot reproduce the problem simulating attached .mos script. It runs fine under the current nightly, am i missing something?

I'm assuming he's using -d=evaluateAllParameters as usual. I will fix it.

[Karim.Abdelhak]

Replying to perost:

Replying to Karim.Abdelhak:

I cannot reproduce the problem simulating attached .mos script. It runs fine under the current nightly, am i missing something?

I'm assuming he's using -d=evaluateAllParameters as usual. I will fix it.

Yes that seems to be the case. Thanks!

[perost]

Should be fixed in ​PR #729.

[perost]

Fixed in b6a7efde.

[casella]

For the record, I was using OMEdit v1.16.0-dev-170-g3af6123f8, and indeed I had the "Evaluate all parameters" checkbox checked.

Unless I really want to re-simulate a model with different parameters, it really makes no sense not to do it: the compilation process is faster, and the simulation process is faster and more numerically robust. Debugging could be a bit harder, because literal constants do not reveal their origin, but unless the model fails I don't really care.

Thanks @perost!