dbstop if error not working

I would like matlab to enter debug mode when there is an error by calling dbstop if error, but it's not working.

In the following code for example, Matlab errors as expected but it doesn't enter debug mode. Are there any settings I should check? I don't see a debug menu.

dbstop if error

ferror123()

function ferror123() % this function errors
    x  = [];
    x(4);
end

[Matt J]

What happens if you use the Pause On Errors menu option instead of dbstop?

[Dylan]

That option doesn't seem to have any effect

Answer 1

[the cyclist]

ferror is the name of a built-in MATLAB function (which the debugger cannot open into). I expect it is being called, rather than your function.

Try renaming your function, and I expect the debugger will be triggered.

[Dylan]

Unfortunately that's not the problem, it's also not stopping on other custom functions I have. And I didn't know ferror was a built-in. I changed the name in this snippet to make it clear that's not the problem

[the cyclist]

Hm. I changed the name of your function to frerror, and it triggered the debugger when I tried to run it.

So, to clarify, you now you have

function ferror123() % this function errors
    x  = [];
    x(4);
end

stored in a file name ferror123.m and you then run

dbstop if error
ferror123

?

Maybe try

dbclear all
dbstop if error

Then I would escalate to restarting MATLAB.

Then I would consider reinstalling MATLAB.

[Dylan]

I don't have it stored as a standalone file, it's just defined within my code. Is that required for dbstop to work? Is there any way to have dbstop work in funcitons defined within the file I'm working in?

Update: I tried defining the function within a file ferror123.m and I still have the same problem

[the cyclist]

I'm pretty confused about how you are defining your function, and how you calling your function.

Let me try to be crystal clear about what I am doing to try and replicate your problem. I am putting the following code into a script file called dbstopError.m

dbstop if error

ferror123()

function ferror123() % this function errors
    x  = [];
    x(4);
end

I have attached that file, for convenience.

Then, I execute that script from the Command Window:

>> dbstopTest

When I do that, MATLAB does enter debug mode, and my Editor window looks like this.

This is all as I would expect.

Can you try that same test?

[Dylan]

That actually does enter debug mode, but I typically run cells of code (split off by lines of %%) by using cmd + enter and that method is still not entering debug mode when I run your file. Doing "Run all sections" (either by clicking the green play button or by pressing cmd + option + enter) does enter debug mode.

So the problem seems to be specifically with command + enter. Is this expected behavior?

[Walter Roberson]

Yes, it is expected behavior for cell mode. cell mode extracts the text of the cell into a temporary .m file and runs that file as a script, but the logic needs to catch errors in order to be able to patch up the line numbers of any errors to refer to the original line numbers instead of to the line numbers relative to the extracted file.

[Dylan]

Ok I see. And just to confirm dbstop if caught error is expected to work in cell mode? That does enter debug mode for me when I do cmd + enter.

[Walter Roberson]

Yes, dbstop if caught error is expected to work in cell mode -- but any line numbers reported might look strange.

Answer 2

[Walter Roberson]

if you are within a try/catch then use

dbstop if caught error

[Dylan]

That did work although I'm not using try/catch. How does that differ from dbstop if error?

[Walter Roberson]

dbstop if error is only for unhandled errors when no try/catch is in effect.

dbstop if caught error is for all error cases... including "expected" errors.

What I mean by expected errors is that in order to deal with compatibility issues or with the possibility of hardware that might or might not be present, it is not uncommon for internal routines to use try/catch protecting a statement that is needed if conditions are right but can be skipped if they are not right.

As an example sometimes it is easier to code

try
  delete(OldObject);
catch ME
end

rather than to code testing to see if OldObject exists and is in a deletable state

[Dylan]

Thanks.

Any reason why dbstop if caught error would work and not dbstop if error?