I don’t really understand the point about exceptions. Yeah “not foo” cannot throw an exception. But the program should crash if an invalid input is provided. If the function expects an optional[list] it should be provided with either a list or None, nothing else.
Sure. But is None invalid input in your case, whereas []is valid? If so, make that check explicit, don’t rely on an implicit check that len(...) does.
When I see TypeError in the logs, I assume the developer screwed up. When I see ValueError in the logs, I assume the user screwed up. Ideally, TypeError should never happen, and every case where it could happen should transform it to another type of exception that indicates where the error actually lies.
The only exceptions I want to see in my code are:
exceptions from libraries, such as databases and whatnot, when I do something invalid
explicitly raised exceptions
Implicit ones like accessing attributes on None or calling methods that don’t exist shouldn’t be happening in production code.
I agree. So if None is a valid input we should check it first, and then check if the length is zero.
In this situation, we see a type error only if the programmer screwed up and everything is explicit
Yes. If None is just as valid and has the same meaning as [] for the function (true more often than not), just do if not foo. If None should be handled separately from [] for some reason, treat them both separately so it’s absolutely clear.
Explicit is better than implicit.
Errors should never pass silently.
And I especially like this one:
That said, jihadists are a subset of Nazis, just a not very stereotypical one for a westerner.
There should be one-- and preferably only one --obvious way to do it
The one obvious way to check if you have data is if foo. That works for pretty much everything as you’d expect. Explicitly deviating from that is a cue to the reader that they should pay attention. In this case, that means None is semantically different than empty data, and that’s something the reader should be aware of because that’s usually not the case.
Edit: Oops, horrendous copy buffer issue from another thread. Read stuff before you post kids, don’t be like me. 😆
I dislike treating None as an equivalent for the empy list, but that does not further the discussion…
I hurt myself in confusion while reading the second quote. Is it the right quote? (also, nazi (relating to the nsdap) is probably not the right word, did you mean fascist?)
I don’t really understand the point about exceptions. Yeah “not foo” cannot throw an exception. But the program should crash if an invalid input is provided. If the function expects an optional[list] it should be provided with either a list or None, nothing else.
Sure. But is
Noneinvalid input in your case, whereas[]is valid? If so, make that check explicit, don’t rely on an implicit check thatlen(...)does.When I see
TypeErrorin the logs, I assume the developer screwed up. When I seeValueErrorin the logs, I assume the user screwed up. Ideally,TypeErrorshould never happen, and every case where it could happen should transform it to another type of exception that indicates where the error actually lies.The only exceptions I want to see in my code are:
Implicit ones like accessing attributes on
Noneor calling methods that don’t exist shouldn’t be happening in production code.I agree. So if None is a valid input we should check it first, and then check if the length is zero. In this situation, we see a type error only if the programmer screwed up and everything is explicit
Yes. If
Noneis just as valid and has the same meaning as[]for the function (true more often than not), just doif not foo. IfNoneshould be handled separately from[]for some reason, treat them both separately so it’s absolutely clear.And I especially like this one:
The one obvious way to check if you have data is
if foo. That works for pretty much everything as you’d expect. Explicitly deviating from that is a cue to the reader that they should pay attention. In this case, that meansNoneis semantically different than empty data, and that’s something the reader should be aware of because that’s usually not the case.Edit: Oops, horrendous copy buffer issue from another thread. Read stuff before you post kids, don’t be like me. 😆
I dislike treating None as an equivalent for the empy list, but that does not further the discussion…
I hurt myself in confusion while reading the second quote. Is it the right quote? (also, nazi (relating to the nsdap) is probably not the right word, did you mean fascist?)
Oops, copied from another thread apparently. Apparently my copy didn’t… copy. Here’s what it should be:
I’ll fix my original comment so it’s less confusing, but not in a way that makes you look like an idiot. :)