Add clamp() function to builtins

[TheComet]

BPO	36788
Nosy	@mdickinson, @stevendaprano, @isidentical, @TheComet

^{Note: these values reflect the state of the issue at the time it was migrated and might not reflect the current state.}

Show more details

GitHub fields:

assignee = None
closed_at = <Date 2019-12-22.10:24:44.416>
created_at = <Date 2019-05-03.21:50:00.429>
labels = ['3.8', 'type-feature', 'library']
title = 'Add clamp() function to builtins'
updated_at = <Date 2020-07-28.07:35:51.947>
user = 'https://github.com/TheComet'

bugs.python.org fields:

activity = <Date 2020-07-28.07:35:51.947>
actor = 'mark.dickinson'
assignee = 'none'
closed = True
closed_date = <Date 2019-12-22.10:24:44.416>
closer = 'mark.dickinson'
components = ['Library (Lib)']
creation = <Date 2019-05-03.21:50:00.429>
creator = 'TheComet'
dependencies = []
files = []
hgrepos = []
issue_num = 36788
keywords = []
message_count = 7.0
messages = ['341358', '341359', '341361', '358780', '358789', '374473', '374474']
nosy_count = 4.0
nosy_names = ['mark.dickinson', 'steven.daprano', 'BTaskaya', 'TheComet']
pr_nums = []
priority = 'normal'
resolution = 'rejected'
stage = 'resolved'
status = 'closed'
superseder = None
type = 'enhancement'
url = 'https://bugs.python.org/issue36788'
versions = ['Python 3.8']

[TheComet]

It would be nice to have a clamp() builtin in addition to min() and max() so one can type e.g. "clamp(value, 2, 5)" instead of having to type "min(max(value, 5), 2)".

[TheComet]

I have implemented it on my branch here: https://github.com/TheComet/cpython/blob/fix-issue-36788/Python/bltinmodule.c

It still needs further testing and I haven't leak checked it properly.

[stevendaprano]

I doubt this is important enough to go into builtins, the only practical use-case I know of for this function is with numbers, so this could go in the math module.

But putting that aside, there are some other problems:

• it isn't clear that clamp() is meaningful for anything that could possibly need a key function;

• the behaviour you have for iterable arguments is inconsistent with the existing behaviour of min(max(x, a), b):

min(max('a string', 'd'), 'm')
=> returns 'd' not ['d', 'd', 'm', 'm', 'm', 'i', 'm', 'g']

• your iterable behaviour is easily done with a comprehension and doesn't need to be supported by the function itself

[clamp(x, a, b) for x in values]

• what do you intend clamp() to do with NAN arguments?

• for numbers, it is sometimes useful to do one-sided clamping, e.g. clamp(x, -1, ∞).

You should read over this thread here:

https://mail.python.org/pipermail/python-ideas/2016-July/041262.html

[isidentical]

Should this issue stay? Maybe port this discussion over python-ideas and after the resolution open it again?

[mdickinson]

Maybe port this discussion over python-ideas and after the resolution open it again?

Sounds good to me. Adding a new builtin function is a big deal that would likely need a PEP. I'm not keen on adding something like this to the math module, either, when it's so simple to construct from max and min.

Closing as rejected here; it can be re-opened if there's a strong consensus for adding it on the ideas list.

[mdickinson]

bpo-41408 was closed as a duplicate of this issue

See also the recent discussion on python-ideas at https://mail.python.org/archives/list/python-ideas@python.org/thread/KWAOQFSV3YJYQV2Y5JXGXFCXHJ3WFLRS/#KWAOQFSV3YJYQV2Y5JXGXFCXHJ3WFLRS

That discussion has petered out without a clear consensus. The next step would likely be for an interested party to put together a PEP.

[mdickinson]

Another python-ideas thread, from 2016: https://mail.python.org/archives/list/python-ideas@python.org/thread/EHZAXE4P2VOT3CE4H6SNDPDASW7H2CGS/