Names in function call exception should have class names, if they're methods

[smurfix]

BPO	2786
Nosy	@gvanrossum, @ncoghlan, @abalkin, @pitrou, @rbtcollins, @ezio-melotti, @smurfix, @vadmium, @serhiy-storchaka, @o11c, @BlckKnght, @refi64, @iritkatriel
Files	full_names.patch: Patch for full method names in exceptions full_names.patch: Patch for full method names in exceptions issue-2786-1.patch

^{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 2021-11-23.17:04:47.532>
created_at = <Date 2008-05-07.20:25:57.605>
labels = ['interpreter-core', 'easy', 'type-feature']
title = "Names in function call exception should have class names, if they're methods"
updated_at = <Date 2021-11-23.17:04:47.531>
user = 'https://github.com/smurfix'

bugs.python.org fields:

activity = <Date 2021-11-23.17:04:47.531>
actor = 'iritkatriel'
assignee = 'none'
closed = True
closed_date = <Date 2021-11-23.17:04:47.532>
closer = 'iritkatriel'
components = ['Interpreter Core']
creation = <Date 2008-05-07.20:25:57.605>
creator = 'smurfix'
dependencies = []
files = ['37994', '38904', '40250']
hgrepos = []
issue_num = 2786
keywords = ['patch', 'easy']
message_count = 28.0
messages = ['66373', '66433', '184151', '184439', '184824', '235314', '235315', '235462', '235476', '238772', '240546', '249081', '249084', '249087', '249174', '249175', '249176', '249178', '249185', '249186', '249188', '249193', '249194', '249268', '249310', '249558', '272016', '406862']
nosy_count = 16.0
nosy_names = ['gvanrossum', 'ncoghlan', 'belopolsky', 'pitrou', 'rbcollins', 'ezio.melotti', 'smurfix', 'Arfrever', 'martin.panter', 'serhiy.storchaka', 'o11c', 'ilblackdragon', 'Steven.Barker', 'refi64', 'xonatius', 'iritkatriel']
pr_nums = []
priority = 'low'
resolution = 'out of date'
stage = 'resolved'
status = 'closed'
superseder = None
type = 'enhancement'
url = 'https://bugs.python.org/issue2786'
versions = ['Python 3.6']

[smurfix]

Consider this simple error:

>>> class foo(object):
...   def __init__(self,bar):
...    pass
... 
>>> foo()
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
TypeError: __init__() takes exactly 2 positional arguments (1 given)
>>> 

The problem is that if that "foo" call is through a variable (or
anything else that obscures which class I'm actually calling) there's no
good way to figure this from the traceback.

The last line should read

TypeError: foo.__init__() takes exactly 2 positional arguments (1 given)

or similar.

[abalkin]

This is similar to bpo-2516.

[ezio-melotti]

Perhaps __qualname__ could be used in the traceback.

[ilblackdragon]

Talked with David Murray (r.david.murray) at @pycon2013 sprints - will try to address this.

[ilblackdragon]

The issue is not that easy to address - because PyFunctionObject is not available from PyEval_EvalCodeEx.

Another note, is that the issue of reporting only function name without class name is observed in many places, not just for example above.

Still, will try to make a patch that will address the issue.

[xonatius]

Made a straightforward patch for this. Probably not that pretty, so suggestions are welcome.

Note that some function names will become pretty long in exceptions:

>>> class A:
...     def __init__(self):
...         def f():
...             pass
...         f(1,2,3)
...
>>> A()
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "<stdin>", line 5, in __init__
TypeError: A.__init__.<locals>.f() takes 0 positional arguments but 3 were given

Passing UTs from lastest snapshot.

[rbtcollins]

I wonder if you could add this to the new code in http://bugs.python.org/issue17911 which I'm hoping to commit this week.

[xonatius]

As long as you think It fits into your issue, I'm ok with adding this patch to it (: it's different files, so no conflicts. Should I just add this patch to your ticket?

[smurfix]

Please do.

[xonatius]

bpo-17911 was submitted. I pulled latest updates and rechecked that "./python -m test -uall" passing with the same patch.