I have a script that performs a bunch of experiments on different datasets, below is part of the logic to determine whether a dataset needs to be loaded. Basically we load the dataset if there is none already loaded, or if the currently loaded one doesn't match the one we need by name.
This example crashes with the error The class dataset has no property or method named 'name' at the if statement (the class does in fact have this property):
if(~exist('dataset','var')||~strcmp(dataset.name,datasets{datai}.id))
loaded=load(datasets{datai}.filename);
dataset=loaded.dataset;
end
If I debug and stop at the line, I can access dataset.name in the debugger without performing any further actions. I don't think the reason is the dataset object not existing. In the loop I ran, the first dataset was correctly loaded, but the second one (where the name check comes into play) wasn't.
This rewriting works:
if(~exist('dataset','var')||~strcmp(nom,datasets{datai}.id))
loaded=load(datasets{datai}.filename);
dataset=loaded.dataset;
nom=dataset.name;
end
Why was I able to access dataset.name in the debugger, and why does the rewriting fix the issue?
Related
I wrote the following custom qgsfunction:
#qgsfunction(args='auto', group='Custom', referenced_columns=[])
def f1(currentfeature, feature, parent):
return "%s\n%s" % (repr(currentfeature),repr(feature))
In the config for the label I enter the following expression:
f1($currentfeature)
And here is the result:
I always believed that the $currentfeature would hold the same value as the implicitly passed argument feature, but they appear to be separate objects.
Can someone please either
confirm that they should be the same value, or
explain the difference between the two
(I also experience other odd things when executing my own custom functions, and I suspect something is not behaving as it should. Sometimes the addresss of the object changes when I zoom in or out, the address seems to cycle between af few values, and simply takes the next value in the cycle when I do any kind of zoom action. And the reason why I got to experiment with this in the first place is because I did not get the expected/correct attribute values back)
My DES model was working well till I've added a new function (just like many others added before) and the model gave me all these error messages:
I wonder what do these messages mean and what made all of them appear suddenly. I was running the model just before it without any errors. Is there a way I could get my model back? :)
Thank you.
Edit 1:
Here is the newly created function: Newly created function
Given that I've ignored it, and made all the callings of it as comments, then the errors were still appearing. I've closed this model version and opened a backup then created the same function and the whole scenario happened again with the same 94 errors.
Is there something I need to change in the function itself?
The function checks the number of agents in certain parts of the model to stop delay blocks accordingly. That's done by adding variables that increment on exiting of agents and get the difference between them.
Typically this is caused by a syntax error in the new function that you created. See if you ignore this and then compile if it gets fixed.
Most of the time the compiler is smart enough to show you the error, but often if you replace a { with [ or you leave out a { the compiler get very confused and gives you 100s of errors.
If yous still cant find the issue maybe paste the code of your new function in the question.
I have linked many parameters in my model to values in the internal ModelDatabase. This works great.
Now I want to allow the user to import a new "Input File" for a specific scenario. I have added a fileChooser element on the Simulation Experiment screen and in the onUpload action I use the ModelDatabase.importFromExternalDB to upload the relevant sheet to the relevant table.
However, this does not seem to work. Initially I thought that the update simply did not happen but when I stop the model in the AnyLogic IDE and start it again in the IDE, the new values are used.
It appears as if the update only happens on startup/close and that the database is "static" while running. I did set the auto-commit parameter to true on the importFromExternalDB function, but this made no difference.
Is there a function I can call to force a "refresh" of the internal database?
You probably load the data at runtime using the default code such as selectFrom("some String with query")
However, these always load from the cached dbase to speed things up.
Instead, you must force AnyLogic to load from the live dbase using an additional argument. In this example, it would be selectFrom(false, "some String query")
This is also documented in the help:
So go through all queries in your model and add the appropriate boolean arg to force-load from the non-cached dbase.
I have a series of ToFile blocks in my Simulink model that each have a unique filename (e.g. "Pulse.mat". I want to store the results of my simulation in timestamped folders based roughly on when I hit the run button / use the sim command.
My solution was to write two scripts, one called during the InitFcn callback and one for the StopFcn/CloseFcn callback (and PreSaveFcn callback).
The InitFcn callback would find all ToFile blocks and change the filename from "Pulse.mat" to something like "../runs//Pulse.mat", and the StopFcn/CloseFcn/PreSaveFcn to revert them to their original ".mat" names. This worked on my small test model, however when I attempted to integrate it into my actual model I receive the following error:
Error evaluating 'StopFcn' callback of block_diagram '<model_name>'.
Caused by:
Cannot change parameter 'Filename' of '<ToFile Block>' while simulation is running. The block was made virtual as it was optimized for simulation
From my research I discovered that the StopFcn actually executes before the simulation is "done" (for whatever reason), but why did it work in my initial test model? Do my ToFile blocks in my actual model have some property set that causes this error to occur?
The block has been virtualised in your larger model; which seems to be changing the execution point. One option might be to untick 'Block Reduction' in the 'Optimisation' pane of the model configuration parameters dialogue.
Alternatively, there might be another approach that you could try -
Setting the model properties 'StartFcn' to something like :
evalin('base','resultTimTag = datestr(clock, ''yyyymmdd_HHhMM'');')
and setting the model properties 'StopFcn' to :
targetDir = evalin('base','resultTimTag');
mkdir(targetDir);
outputs = dir('*.mat');
for i=1:length(outputs)
movefile(outputs(i).name,targetDir);
end
evalin('base','clear resultTimeTag');
It's possibly a little more brute-force than your approach, but seems to work quite nicely...
I have got some inputs in my model that are passed through the environment controller block. In this way, I can use test inputs (fixed or configurable parameters) in simulation and the top-level input ports for Embedded Code Generation. However, I need to check the outputs of my model (after simulation) on a scope.
I was hoping that there is way I can use something similar to environment controller block in order to ignore scope blocks when I generate code. In this way, I can still view my outputs but not worry about any unnecessary blocks getting error-flagged during code generation checking. Is there any way to do it or is it "Automatically" ignored when generating code. Has anyone come across this?
KR,