Repository navigation
io class name are hardcoded in reprs #66060
Description
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 = None
created_at = <Date 2014-06-24.17:36:58.710>
labels = ['extension-modules', 'easy', 'type-bug', 'expert-IO']
title = 'io class name are hardcoded in reprs'
updated_at = <Date 2022-01-23.10:01:52.325>
user = 'https://github.com/serhiy-storchaka'bugs.python.org fields:
activity = <Date 2022-01-23.10:01:52.325>
actor = 'asaka'
assignee = 'none'
closed = False
closed_date = None
closer = None
components = ['Extension Modules', 'IO']
creation = <Date 2014-06-24.17:36:58.710>
creator = 'serhiy.storchaka'
dependencies = []
files = []
hgrepos = []
issue_num = 21861
keywords = ['patch', 'easy']
message_count = 6.0
messages = ['221476', '221990', '221993', '222011', '222939', '342684']
nosy_count = 12.0
nosy_names = ['rhettinger', 'pitrou', 'benjamin.peterson', 'stutzbach', 'Claudiu.Popa', 'hynek', 'piotr.dobrogost', 'serhiy.storchaka', 'josh.r', 'asaka', 'Windson Yang', 'Philip Dye']
pr_nums = ['14774', '30824']
priority = 'normal'
resolution = None
stage = 'patch review'
status = 'open'
superseder = None
type = 'behavior'
url = 'https://bugs.python.org/issue21861'
versions = ['Python 2.7', 'Python 3.3', 'Python 3.4']
Activity
serhiy-storchaka commentedon Jun 24, 2014
Expected results are "<main.F name='/dev/null' mode='rb'>", "<main.B name='/dev/null'>" and "<main.T name='/dev/null' encoding='UTF-8'>". Usually reprs of subclass instances substitute actual module and class names.
PCManticore commentedon Jun 30, 2014
The same should be done for _pyio?
MojoVampire commentedon Jun 30, 2014
Is it common for C implementations to introspect to figure out their "real" name? I do this manually for reprs of my user defined classes, but I haven't noticed many built-ins that consider extensibility for the repr. Maybe I'm just not using the classes that do it or I'm overriding the repr without checking?
I just tested, and it looks like frozenset has a subclass friendly repr, while bytearray does not as of 3.4.0. Seems like it might make sense to first determine if introspection should be the default; it would mean a little code bloat every time it's done.
rhettinger commentedon Jul 1, 2014
This is a sensible idea that makes life easier for people writing subclasses.
+1 for your suggestion.
serhiy-storchaka commentedon Jul 13, 2014
I now see that this issue is not so easy from C side.
And more, it is very uncommon for either Python or C implementations (especial C implementations) to introspect to figure out their "real" name. Existing cases are rather exceptions. And in many case this is only a side effect of sharing an implementation between several classes (set and frozenset, three buffered file classes in io, etc). In many cases repr's flexibility is limited: hardcoded or omitted module name, used class's __name__ instead of __qualname__, etc.
This should be discussed on Python-Dev maillist.
Windsooon commentedon May 17, 2019
IIUC, in the c code we just hardcode the name "_io.FileIO" for "PyFileIO_Type" in https://github.com/python/cpython/blob/master/Modules/_io/fileio.c#L1180. If we want to get a dynamic name, we should replace the hardcode name with the module name (like "__main__"). I found this also happens on other modules like collections.
gh-66060: Use actual class name in _io type's __repr__ (#30824)
pythongh-66060: Use actual class name in _io type's __repr__ (python#…
pythongh-66060: Use actual class name in _io type's __repr__ (python#…
pythongh-66060: Use actual class name in _io type's __repr__ (python#…