STINNER Victor <[email protected]> added the comment:
Bytes program name problem should be splitted in two parts:
(a) subprocess.call([b'env']) and subprocess.call([b'env'], env={'PATH':
'/usr/bin'}): bytes program and unicode environ
(b) bytes program and bytes environ
Part (a)
--------
(a) would benefit from the creation of os(.path).fsencode() function (issue
#8514).
Part (b)
--------
If the creation of os.environb is accepted (#8603), I think that subprocess
should also be modified to support pure bytes environ. Mix bytes and unicode
variables in env would lead to mojibake and headaches.
So I propose the creation of an "envb" (environment bytes) argument to Popen
constructor. (b) would be written as:
subprocess.call([b'env'], envb={b'PATH': b'/usr/bin'})
or
envb = os.environb.copy()
envb[b'PATH'] = b'/usr/bin'
subprocess.call([b'env'], envb=envb)
(and it will not be possible to define env and envb at the same time)
In this case, we will need a pure bytes version of os.get_exec_path(), and
os._execvpe() should have a "bytes environment" mode (add an optional argument).
--
I have a patch fixing both problems, parts (a) and (b), but the patch depends
on os.environb. I prefer to wait until os.environb is accepted to submit my
patch.
----------
dependencies: +Create a bytes version of os.environ and getenvb() -Create
fsencode() and fsdecode() functions in os.path
_______________________________________
Python tracker <[email protected]>
<http://bugs.python.org/issue8513>
_______________________________________
_______________________________________________
Python-bugs-list mailing list
Unsubscribe:
http://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com
