bug-gnulib@gnu.org mirror (unofficial)
 help / color / mirror / Atom feed
From: Bruno Haible <bruno@clisp.org>
To: bug-gnulib@gnu.org
Cc: Dima Pasechnik <dimpase@cs.ox.ac.uk>,
	Collin Funk <collin.funk1@gmail.com>
Subject: Re: Python != None
Date: Sun, 25 Feb 2024 12:57:15 +0100	[thread overview]
Message-ID: <41613867.J2Yia2DhmK@nimes> (raw)
In-Reply-To: <005ECEB2-5D0D-43C6-8AA2-B44D6AF0219E@cs.ox.ac.uk>

Collin Funk wrote:
> One thing that annoys me personally
> is comparing to none using "!=" instead of "is not". This is
> recommended against in PEP 8, "Comparisons to singletons like None
> should always be done with is or is not, never the equality
> operators." [1].

Dima Pasechnik wrote:
> The pythonic way is
> 
>     mode is not None
> 
> rather than 
> 
>     mode != None
> 
> (the reason is None is an object)

Summarizing the arguments for "is not None":
  1- Clarity and readability, best practice, PEP 8.
  2- Identity comparison vs. value comparison.
  3- None is a singleton.
  4- Prevents wrong results if an __eq__ method has been incorrectly coded.

Here's my take on it.

1- is clearly subjective. Clarity is context dependent. Etc.
2- Yes, https://docs.python.org/3.12/reference/expressions.html#comparisons
   clearly explains the difference between == and 'is'.
   (It's more or less like the difference between EQUAL and EQ in Lisp.)
3- None is a singleton, but it nonetheless supports both == and 'is'. So,
   this is a weak argument.
4- This is one of the weakest possible arguments.

The style warnings about "!= None" in pycodestyle and/or pylint are
relativized by this warning in Python itself:

  >>> '' != None
  True
  >>> '' is not None
  <stdin>:1: SyntaxWarning: "is not" with a literal. Did you mean "!="?
  True

So, if Python itself warns about some code that follows PEP 8...
it means that we need to think carefully, rather than blindly apply
that PEP 8 rule.

I think a better rule, in particular in the context of gnulib-tool.py,
is to observe that each variable has a certain set of possible values
(even though this set of values is not explicitly stated in the source
code), and with this set of possible values comes a comparison operator.

So, what I mean is:

  * If a variable can only contain objects like GLModule instances or None,
    then 'is' (pointer comparison) is the right choice for this variable.

  * If a variable can only contain strings or None, then '==' (value
    comparison) is the right choice for this variable, since comparing
    strings with 'is' is unreliable:

      >>> x = "ab"
      >>> y = "ab"
      >>> x is y
      True
      
      >>> x = "abc"[:2]
      >>> y = "abx"[:2]
      >>> x is y
      False

  * If a variable can only contain a list or None, then '==' (value comparison)
    is the right choice for this variable.

    >>> ["a"] == ["a"]
    True
    >>> ["a"] is ["a"]
    False

So, depending on the variable:
   mode != None                       OK
   modules != None                    OK
   module != None                     not OK, better write:   module is not None

Bruno





  reply	other threads:[~2024-02-25 11:57 UTC|newest]

Thread overview: 28+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2024-02-23  5:23 [PATCH] gnulib-tool.py: Follow gnulib-tool changes, part 27 Collin Funk
2024-02-23 13:08 ` Bruno Haible
2024-02-23 22:20   ` Collin Funk
2024-02-23 23:51     ` Bruno Haible
2024-02-24  2:36       ` Collin Funk
2024-02-24  5:49         ` gnulib-tool.py: Follow gnulib-tool changes, part 28 Collin Funk
2024-02-24 23:25           ` gnulib-tool.py: Follow gnulib-tool changes, part 27 Bruno Haible
2024-02-25  0:03             ` Dima Pasechnik
2024-02-25 11:57               ` Bruno Haible [this message]
2024-02-25 19:29                 ` Python != None Collin Funk
2024-02-25 20:07                   ` Collin Funk
2024-02-26 20:38                     ` pycodestyle configuration Bruno Haible
2024-02-26 21:31                       ` Collin Funk
2024-02-26 22:54                         ` Bruno Haible
2024-02-27  0:51                           ` Collin Funk
2024-02-27  2:38                             ` Bruno Haible
2024-02-27  4:22                               ` Collin Funk
2024-02-25 20:55                   ` Python != None Dima Pasechnik
2024-02-25 12:02             ` Python 'strings' Bruno Haible
2024-02-25 19:05               ` Collin Funk
2024-02-24 23:42           ` gnulib-tool.py: Follow gnulib-tool changes, part 28 Bruno Haible
2024-02-25  0:47             ` Collin Funk
2024-02-25  1:18               ` Collin Funk
2024-02-25  1:25                 ` Bruno Haible
2024-02-25  3:32                   ` Collin Funk
2024-02-26 20:51                     ` Bruno Haible
2024-02-28 11:51                       ` Collin Funk
2024-02-28 12:14                         ` Bruno Haible

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

  List information: https://lists.gnu.org/mailman/listinfo/bug-gnulib

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=41613867.J2Yia2DhmK@nimes \
    --to=bruno@clisp.org \
    --cc=bug-gnulib@gnu.org \
    --cc=collin.funk1@gmail.com \
    --cc=dimpase@cs.ox.ac.uk \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).