Inconsistent sys.path between python and pdb

[hexagonrecursion]

BPO	42384
Nosy	@gvanrossum, @miss-islington, @hexagonrecursion
PRs	bpo-42384: pdb: correctly populate sys.path[0] #23338 [3.9] bpo-42384: pdb: correctly populate sys.path[0] (GH-23338) #24290 [3.8] bpo-42384: pdb: correctly populate sys.path[0] (GH-23338) #24291 [3.8] bpo-42384: pdb: correctly populate sys.path[0] (GH-23338) #24320 [3.9] bpo-42384: pdb: correctly populate sys.path[0] (GH-23338) #24321 bpo-42383: pdb: do not fail to restart the target if the current directory changed #23412

^{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 2021-01-26.15:59:07.336>
created_at = <Date 2020-11-17.07:46:41.774>
labels = []
title = 'Inconsistent sys.path between python and pdb'
updated_at = <Date 2021-01-26.15:59:07.335>
user = 'https://github.com/hexagonrecursion'

bugs.python.org fields:

activity = <Date 2021-01-26.15:59:07.335>
actor = 'gvanrossum'
assignee = 'none'
closed = True
closed_date = <Date 2021-01-26.15:59:07.336>
closer = 'gvanrossum'
components = []
creation = <Date 2020-11-17.07:46:41.774>
creator = 'hexagonrecursion'
dependencies = []
files = []
hgrepos = []
issue_num = 42384
keywords = ['patch']
message_count = 8.0
messages = ['381215', '381218', '381219', '385464', '385470', '385482', '385666', '385667']
nosy_count = 3.0
nosy_names = ['gvanrossum', 'miss-islington', 'hexagonrecursion']
pr_nums = ['23338', '24290', '24291', '24320', '24321', '23412']
priority = 'normal'
resolution = 'fixed'
stage = 'resolved'
status = 'closed'
superseder = None
type = None
url = 'https://bugs.python.org/issue42384'
versions = []

[hexagonrecursion]

The first entry in sys.path is different between python foo.py and python -m pdb foo.py. In the former it is the absolute path to the parent directory of foo.py while in the later it is a relative path (unless the debug target was specified using an absolute path). The meaning of the absolute path does not change when the current directory changes (e.g. via os.chdir()) while the meaning of the relative path does. Like any environment inconsistency between regular program execution and the debugger this may lead to bugs that mysteriously vanish when you try to debug them.

$ cat > print-path.py
import sys
from pprint import pprint

pprint(sys.path)
$ python3 print-path.py 
['/home/user',
 '/usr/lib64/python38.zip',
 '/usr/lib64/python3.8',
 '/usr/lib64/python3.8/lib-dynload',
 '/usr/lib64/python3.8/site-packages',
 '/usr/lib/python3.8/site-packages']
$ python3 -m pdb print-path.py                                  
> /home/user/print-path.py(1)<module>()
-> import sys
(Pdb) c
['',
 '/usr/lib64/python38.zip',
 '/usr/lib64/python3.8',
 '/usr/lib64/python3.8/lib-dynload',
 '/usr/lib64/python3.8/site-packages',
 '/usr/lib/python3.8/site-packages']

[hexagonrecursion]

I'll look into fixing this after I fix bpo-42383

[hexagonrecursion]

After reading Lib/pdb.py:main I believe I can fix both by changing a single line.

[miss-islington]

New changeset 8603dfb by Andrey Bienkowski in branch 'master':
bpo-42384: pdb: correctly populate sys.path[0] (GH-23338)
8603dfb

[gvanrossum]

Hey Andrey, this has been merged into 3.10, but the backports didn't work because the structure of the tests has changed (os_helper doesn't exist). Do you want to fix these?

[hexagonrecursion]

I'll give it a try

[gvanrossum]

New changeset f2df795 by Andrey Bienkowski in branch '3.9':
[3.9] bpo-42384: pdb: correctly populate sys.path[0] (GH-23338) (bpo-24321)
f2df795

[gvanrossum]

New changeset c10180e by Andrey Bienkowski in branch '3.8':
[3.8] bpo-42384: pdb: correctly populate sys.path[0] (GH-23338) (bpo-24320)
c10180e