LWN: Comments on "Corner cases and exception types"

Corner cases and exception types

nevyn — Sun, 25 Aug 2019 22:43:11 +0000

So how would you rewrite the "filtered_data" example? It's plenty concise for me.

Which is:

    filtered_data = [y for x in data if (y := f(x)) is not None]

...the obvious simple choice is:

    # Just iterate data and return the value we want...
    filtered_data = [x for x in iter_f(data) if x is not None]

...which leads to the even better version:

    # Get a list of filtered values we want...
    filtered_data = list_filter_f(data)

...the original idea behind python wasn't to cram everything on a single line to win perl golf competitions.

Corner cases and exception types

nybble41 — Thu, 22 Aug 2019 16:34:38 +0000

filtered_data = [y for y in map(f, data) if y is not None]

Corner cases and exception types

smurf — Wed, 21 Aug 2019 20:27:38 +0000

So how would you rewrite the "filtered_data" example? It's plenty concise for me.

Corner cases and exception types

robert_s — Sun, 18 Aug 2019 12:12:55 +0000

While I can see its use in `if` statements, its use in comprehensions seems to encourage an extremely confusing approach, mutating state in the middle of comprehensions. The power and clarity of comprehension statements comes from their use as functional, pure constructs. All of the examples I've seen of the walrus operator used in comprehensions are confusing at best.

Corner cases and exception types

nivedita76 — Wed, 14 Aug 2019 16:03:37 +0000

The instances that are being turned into errors don’t make sense. It’s not banning := inside a comprehension, it’s banning it when you try to overwrite an iteration variable with it.

Corner cases and exception types

nivedita76 — Wed, 14 Aug 2019 16:01:52 +0000

The leaking out is intentional. The PEP gives examples where it’s useful.

Corner cases and exception types

rsidd — Wed, 14 Aug 2019 09:12:49 +0000

[replying to myself, sorry] However, this leaking would seem to be a bug in the implementation of := at the language level, not a bug by the programmer. As I understand it, it is very hard to fix in the cpython reference implementation. But therefore turning it into a programming bug seems wrong, and considerably reduces the usefulness of the operator (IMO).

Corner cases and exception types

agateau — Wed, 14 Aug 2019 09:12:47 +0000

What surprises me here is that the new variable is available in the outer scope. I would expect whatever variable is defined in a comprehension list to stay in the comprehension list.

Corner cases and exception types

rsidd — Wed, 14 Aug 2019 09:08:18 +0000

Interestingly, in python2 the variable inside a comprehension does leak out. This has bitten me a few times. For example, typing "[0 for i in range(5)]" will set the variable i=4 in python2, but not in python3.

Corner cases and exception types

xanni — Wed, 14 Aug 2019 01:56:36 +0000

I disagree. I find that it brings considerable value that more than justifies the added complexity. It's one of the features present in other languages that I have long missed in Python and I'm very pleased to see it added.

Corner cases and exception types

Cyberax — Tue, 13 Aug 2019 21:49:16 +0000

Just remove the walrus operator entirely. It provides close to no improvements but adds complexity.