A public facing API for __unittest = True

[Julian]

BPO	15891
Nosy	@bitdancer, @voidspace, @Julian, @ericsnowcurrently
Superseder	bpo-1705520: API for excluding methods from unittest stack traces

^{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 2012-09-09.19:36:08.318>
created_at = <Date 2012-09-09.19:08:04.375>
labels = ['type-feature', 'library']
title = 'A public facing API for __unittest = True'
updated_at = <Date 2012-09-10.14:32:45.374>
user = 'https://github.com/Julian'

bugs.python.org fields:

activity = <Date 2012-09-10.14:32:45.374>
actor = 'eric.snow'
assignee = 'none'
closed = True
closed_date = <Date 2012-09-09.19:36:08.318>
closer = 'r.david.murray'
components = ['Library (Lib)']
creation = <Date 2012-09-09.19:08:04.375>
creator = 'Julian'
dependencies = []
files = []
hgrepos = []
issue_num = 15891
keywords = []
message_count = 4.0
messages = ['170126', '170128', '170129', '170191']
nosy_count = 4.0
nosy_names = ['r.david.murray', 'michael.foord', 'Julian', 'eric.snow']
pr_nums = []
priority = 'normal'
resolution = None
stage = 'resolved'
status = 'closed'
superseder = '1705520'
type = 'enhancement'
url = 'https://bugs.python.org/issue15891'
versions = ['Python 3.3', 'Python 3.4']

[Julian]

Can __unittest = True, which is used to hide stack frames from tracebacks inside the testing framework, be documented as being public, or, if that's not acceptable, be replaced by something that is? _exc_info_to_string is a hairy function to need to rewrite when authoring a testing framework, but it's a rather common thing to want, and would be nice if it was easily done by integrating with what TestCase is already doing.

It's also (with or without formal blessing) being used in the wild in at least one place I know of here http://bazaar.launchpad.net/~testtools-committers/testtools/trunk/view/head:/testtools/testcase.py#L798 and since it's not terribly critical and will fail rather nicely even if the API changes suddenly I'd bet other places are using it too, though I haven't bothered to check.

See also http://twistedmatrix.com/trac/ticket/4127 which would be easier to implement if this were acceptable to use for external code.

[bitdancer]

This is a duplicate of bpo-1705520. Since Michael obviously hasn't gotten to it yet, a proposed patch might move things along.

[voidspace]

Whatever we do must be backwards compatible. The question is, is a better api that doesn't use stack frames (which is a hack in my opinion) even possible?

[ericsnowcurrently]

yeah, I keep thinking we need a better API for managing tracebacks. One of these days...