Investigate why the Model Editor is not applying the test on Apply

[smk78]

Created at request of @rozyczko :
The Model Editor window does not seem to have the test run implemented on Apply. Was this just missed, or was there a reason?

[butlerpd]

@smk78 and/or @rozyczko, is this referring to the create model editor? or the full model editor? My impression is the the error checking is pretty robust now on the "add custom model" editor. I note that the "edit custom model" editor does not have an apply button, just a save button. However it still seems to catch simple syntax errors. There is another issue #1595 which describes what happens with those that still don't get caught.
Question: can we close this or is there something more specific to look for?

[smk78]

Add Custom Model.

So if you create one:

And click Apply all is good as you'd expect.

Now deliberately introduce an error (so here I'll change the function to have a parameter b but not update the parameter list):

This time when you click apply (and note I've had to check the overwrite box at the top) the syntax checker spots the error.

But if I switch to the Model Editor tab and make the same code change:

And then click Apply the syntax checking is not applied and the faulty model is saved:

[butlerpd]

Interesting .. I had not noticed the different buttons when the edit custom model editor is automatically loaded after a first save vs when it is called directly from the menu bar. It clearly is the same editor code (if not it should be).

That behavior is similar to what I found when the custom model editor is called directly. A deliberate mistake in the parameter table for example (in my case removal of a comma) caused the problem described here (no error flagged and file saved but ..), but changing the math function to have an unclosed parenthesis [ x(x+y(r)] for example DID cause the red box "error" to pop up. As I say, the add custom model panel seems to be much more robust at finding errors, but I suspect making the general python editor that robust may be more problematic? It would be nice to have some ideas of what is possible or not to help the users here I think.

[lucas-wilkins]

@rozyczko Was this investigated? Was it fixed? Do we need an issue that actions a fix? Can we close?