Enable DeprecationWarnings by default when not explicit in unittest.main()

[gpshead]

BPO	39892
Nosy	@gpshead, @ncoghlan, @tirkarthi

^{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 2020-11-08.23:29:02.511>
created_at = <Date 2020-03-08.00:53:52.939>
labels = ['tests', 'type-feature', 'library', '3.9', 'invalid']
title = 'Enable DeprecationWarnings by default when not explicit in unittest.main()'
updated_at = <Date 2020-11-08.23:29:02.510>
user = 'https://github.com/gpshead'

bugs.python.org fields:

activity = <Date 2020-11-08.23:29:02.510>
actor = 'gregory.p.smith'
assignee = 'none'
closed = True
closed_date = <Date 2020-11-08.23:29:02.511>
closer = 'gregory.p.smith'
components = ['Library (Lib)', 'Tests']
creation = <Date 2020-03-08.00:53:52.939>
creator = 'gregory.p.smith'
dependencies = []
files = []
hgrepos = []
issue_num = 39892
keywords = []
message_count = 3.0
messages = ['363620', '364281', '380563']
nosy_count = 3.0
nosy_names = ['gregory.p.smith', 'ncoghlan', 'xtreak']
pr_nums = []
priority = 'normal'
resolution = 'not a bug'
stage = 'resolved'
status = 'closed'
superseder = None
type = 'enhancement'
url = 'https://bugs.python.org/issue39892'
versions = ['Python 3.9']

[gpshead]

Recurring theme: The stdlib has the issue of DeprecationWarning being added to APIs we are changing or removing a few versions in the future yet we perceive that many people never actually bother to try checking their code for deprecation warnings. Only raising issues with a documented, warned, planned in advance API change when it actually happens.

Could we reduce the chances of this by enabling DeprecationWarnings by default for processes started via unittest.main() and other common unittest entrypoints? (other test frameworks like pytest should also consider this if they don't already; do we have any existing external implementations of this for inspiration?)

One issue with this is that some important warnings are at _parse_ time or _import_ time. But we can deal with import time ones if we are able to have the unittest entrypoint re-exec the process with the same args but with warnings enabled. (and _could_ surface parse time ones if we're willing to accept slower process startup by disabling use of pycs; i wouldn't go that far)

Related work: https://www.python.org/dev/peps/pep-0565/

has this idea already been discussed? I don't remember and haven't searched backwards...

[ncoghlan]

bpo-10535 says they should already be on by default in unittest.

I seem to recall checking that was still true when implementing the default warning filter changes in 3.7.

[gpshead]

my bad :)