Repository navigation
Policy for altering sys.path #68323
Description
| BPO | 24135 |
|---|---|
| Nosy | @bitdancer |
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 2015-05-06.13:22:15.320>
created_at = <Date 2015-05-06.13:03:53.199>
labels = ['invalid']
title = 'Policy for altering sys.path'
updated_at = <Date 2015-05-06.13:22:15.318>
user = 'https://bugs.python.org/ThomasGttler'bugs.python.org fields:
activity = <Date 2015-05-06.13:22:15.318>
actor = 'r.david.murray'
assignee = 'none'
closed = True
closed_date = <Date 2015-05-06.13:22:15.320>
closer = 'r.david.murray'
components = []
creation = <Date 2015-05-06.13:03:53.199>
creator = 'Thomas G\xc3\xbcttler'
dependencies = []
files = []
hgrepos = []
issue_num = 24135
keywords = []
message_count = 2.0
messages = ['242664', '242671']
nosy_count = 2.0
nosy_names = ['r.david.murray', 'Thomas G\xc3\xbcttler']
pr_nums = []
priority = 'normal'
resolution = 'not a bug'
stage = 'resolved'
status = 'closed'
superseder = None
type = None
url = 'https://bugs.python.org/issue24135'
versions = []
Activity
ThomasGttler commentedon May 6, 2015
I am missing a policy how sys.path should be altered.
We run a custom sub class of list in sys.path. We set it in sitecustomize.py
This instance get replace by a common list in lines like this:
The above line is from pip, it similar things happen in a lot of packages.
Before trying to solve this with code, I think the python community should agree an a policy for altering sys.path.
What can I do to this done?
We use Python 2.7.
bitdancer commentedon May 6, 2015
My guess, unfortunately, is give up. There is too much code in the field that replaces sys.path with a regular list. What you propose is effectively an impossible social engineering problem. You may well also run into issues where the Python C code calls list methods directly even if sys.path is a subclass of list. (I don't know that that is true, but there are certainly places where the C code does that kind of thing).
Could you get consensus to change stdlib and pip practice? That's at least conceivable. I guess python-ideas would be the place to start for that. I'm going to close the issue, since this would be a PEP level change if you are successful. (Frankly, I don't think you are likely to be successful, but I don't hang out on python-ideas so I might be wrong :)