Using raise for Effective Exceptions (Summary)

Great tutorial and lots of context and practicality here. What hit home for my current project is to only use exceptions to convey a better message than what you get by default. I always wondered why I’m using a custom exception when the default messaging was already clear? I went back through my code and removed my custom exceptions.

Exceptions groups look cool!

Thanks folks, glad you enjoyed the course.

Yep, I find I use built-in exceptions the majority of the time. I only tend to create custom ones if I truly need to express something original.

As for the groups, yeah, they’re neat. I’ve yet to use them in real life, but handy to have when the need arises.

Great tutorial!!

This is reference a script which does the following task.

Function A, calls an API 1 to GET the data 1 - Service 1 Function B, calls an API 2 to GET the data 2 - Service 2 Function C, using the data 1, 2 does a POST to update values - Service 3

The Function A, B, C are being called on a list items.

What will be a good strategy to handle exception in the above scenario?

1. Catching the exception using a try and except at the individual function level or in the main function?
2. What will be the good approach to handle exception being passed between different functions ?

Any pointers are much appreciated?

Hi Gary,

When you’ve got multiple things that can raise exceptions like in your example, the question you need to ask yourself is: what are you going to do if there is an exception? In the example you gave, I suspect, all three things have to work, or nothing is going to work. If, for example, the only thing you want is a pretty error for your users instead of the exception propagating all the way up, then you really only have to catch things at the outside level.

On the other hand, if say you want to do a re-try on one of the API calls, then you might catch the individual exception inside a retry loop, and then re-raise the exception after the 3 (or whatever) re-attempts.

What will be the good approach to handle exception being passed between different functions ?

Ok, I’m going to be pedantic here: exceptions don’t really get “passed between functions”. Exceptions propagate up the stack. So if A calls B and B throws but doesn’t catch, then the calling point of B inside A is the next place a catch can happen. If A doesn’t catch it and nobody has wrapped A, then the program sees it and causes the error out.

I think what you’re asking is “at what level should I catch”, and that really comes back to my “what are you going to do with it when you catch it?”. General advice is catch it as close to the source as possible if you’re going to do something with it.

Say you were going to catch a KeyError because something wasn’t in a dict and if it isn’t there you were going to add something to the dict (there are easier ways of doing this, see deafultdict, but let’s keep the example simple). In this case, you want to wrap the point where you are accessing the dict and expecting the KeyError.

Instead, if you’re doing a bunch of I/O operations and if any of them fail you want to close the file and error out, then you’d put all of them inside the try. You don’t want to write the error handling code over and over.

The extreme of this goes back to my first answer to your question: if all you’re doing is printing an error instead of having the program crash, then the outer most calling point is fine.

Hope this helps.