This issue tracker has been migrated to GitHub, and is currently read-only.
For more information, see the GitHub FAQs in the Python's Developer Guide.

Author aeros
Recipients aeros, docs@python
Date 2020-03-23.00:50:31
SpamBayes Score -1.0
Marked as misclassified Yes
Message-id <1584924633.15.0.655954011008.issue40045@roundup.psfhosted.org>
In-reply-to
Content
In a recent python-ideas thread, the rule of dunder methods being reserved for Python internal usage only was brought up (https://mail.python.org/archives/list/python-ideas@python.org/message/GMRPSSQW3SXNCP4WU7SYDINL67M2WLQI/), due to an author of a third party library using them without knowing better. Steven D'Aprano linked the following section of the docs that defines the rule: https://docs.python.org/3/reference/lexical_analysis.html#reserved-classes-of-identifiers. 

When I had attempted to search for the rule in the documentation (prior to the above discussion), I noticed that it was rather difficult to discover because it was written just as "System-defined names" with no mention of "dunder" (which is what the dev community typically refers to them as, at least in more recent history).

To make it easier for the average user and library maintainer to locate this section, I propose changing the first line to one of the following:

1) System-defined names, also known as "dunder" names.
2) System-defined names, informally known as "dunder" names.

I'm personally in favor of (1), but I could also see a reasonable argument for (2). If we can decide on the wording, it would make for a good first-time PR to the CPython docs.
History
Date User Action Args
2020-03-23 00:50:33aerossetrecipients: + aeros, docs@python
2020-03-23 00:50:33aerossetmessageid: <1584924633.15.0.655954011008.issue40045@roundup.psfhosted.org>
2020-03-23 00:50:33aeroslinkissue40045 messages
2020-03-23 00:50:31aeroscreate