Hexadecimal id in reprs

[serhiy-storchaka]

BPO	22031
Nosy	@vstinner, @stevendaprano, @serhiy-storchaka
Files	repr_hex_id.diff

^{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 = 'https://github.com/serhiy-storchaka'
closed_at = <Date 2014-07-22.09:17:52.341>
created_at = <Date 2014-07-22.08:50:41.985>
labels = ['type-bug', 'library']
title = 'Hexadecimal id in reprs'
updated_at = <Date 2014-07-22.11:52:11.415>
user = 'https://github.com/serhiy-storchaka'

bugs.python.org fields:

activity = <Date 2014-07-22.11:52:11.415>
actor = 'vstinner'
assignee = 'serhiy.storchaka'
closed = True
closed_date = <Date 2014-07-22.09:17:52.341>
closer = 'serhiy.storchaka'
components = ['Library (Lib)']
creation = <Date 2014-07-22.08:50:41.985>
creator = 'serhiy.storchaka'
dependencies = []
files = ['36025']
hgrepos = []
issue_num = 22031
keywords = ['patch']
message_count = 6.0
messages = ['223647', '223648', '223652', '223653', '223660', '223662']
nosy_count = 4.0
nosy_names = ['vstinner', 'steven.daprano', 'python-dev', 'serhiy.storchaka']
pr_nums = []
priority = 'normal'
resolution = 'fixed'
stage = 'resolved'
status = 'closed'
superseder = None
type = 'behavior'
url = 'https://bugs.python.org/issue22031'
versions = ['Python 3.5']

[serhiy-storchaka]

Default repr of an instance looks as:

>>> object()
<object object at 0xb71a14a8>

Many specialized reprs follow a pattern <... at {hexadecimal id}...>. But there are few deviations:

1. Reprs of WeakValueDictionary and WeakKeyDictionary in the weakref module output decimal id.

2. reprlib and reprs of CDLL in the ctypes module and DateTime and MultiCall in the xmlrpc.client module output hexadecimal id but without the "0x" prefix.

Proposed patch makes these cases to conform with other reprs.

[vstinner]

The patch looks good to me.

[python-dev]

New changeset 4cef7b0ec659 by Serhiy Storchaka in branch 'default':
Issue bpo-22031: Reprs now always use hexadecimal format with the "0x" prefix
http://hg.python.org/cpython/rev/4cef7b0ec659

[serhiy-storchaka]

Thanks Victor. Applied only in default because this issue is too minor.

[stevendaprano]

Many specialized reprs follow a pattern <... at {hexadecimal id}...>.
But there are few deviations:
[...]
Proposed patch makes these cases to conform with other reprs.

I oppose making this change. The exact format of this generic repr of
instances is not a part of Python's API, and we should not encourage
people to think that it must always follow the same pattern or rely on
that pattern.

Changing the repr risks breaking doctests. That *might* be acceptable if
there was some obvious benefit to the change, e.g. changing from some
generic <whatever at some_id> to a nicer, more useful display. But to
risk breaking doctests just for consistency with other generic displays
when that consistency has never been promised is not worth it.

[vstinner]

Changing the repr risks breaking doctests.

That's why the change was only done in Python 3.5. To port your application to Python 3.5, you can modify you doctests to use "..." (ellipsis option of doctests).

That *might* be acceptable if there was some obvious benefit
to the change,

It's better to have consistent representation of objects. Low-level debuggers like gdb uses hexadecimal, not decimal. I'm regulary using gdb, and having the memory address in hexadecimal helps.

By the way, you should never rely on the (exact) representation in your unit tests :-)