Repository navigation
Allow setting cell value #74671
Description
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 = 'https://github.com/lisroach'
closed_at = <Date 2017-06-08.11:45:01.119>
created_at = <Date 2017-05-26.16:17:01.425>
labels = ['interpreter-core', 'type-feature', '3.7']
title = 'Allow setting cell value'
updated_at = <Date 2017-09-20.13:54:17.955>
user = 'https://github.com/pitrou'bugs.python.org fields:
activity = <Date 2017-09-20.13:54:17.955>
actor = 'vstinner'
assignee = 'lisroach'
closed = True
closed_date = <Date 2017-06-08.11:45:01.119>
closer = 'serhiy.storchaka'
components = ['Interpreter Core']
creation = <Date 2017-05-26.16:17:01.425>
creator = 'pitrou'
dependencies = []
files = []
hgrepos = []
issue_num = 30486
keywords = []
message_count = 10.0
messages = ['294554', '294573', '294577', '294583', '295057', '295058', '295416', '295417', '295476', '302617']
nosy_count = 9.0
nosy_names = ['rhettinger', 'terry.reedy', 'ncoghlan', 'pitrou', 'vstinner', 'benjamin.peterson', 'serhiy.storchaka', 'lisroach', 'Jim Fasarakis-Hilliard']
pr_nums = ['1840', '3668']
priority = 'normal'
resolution = 'fixed'
stage = 'resolved'
status = 'closed'
superseder = None
type = 'enhancement'
url = 'https://bugs.python.org/issue30486'
versions = ['Python 3.7']
Activity
pitrou commentedon May 26, 2017
There are use cases for setting a cell value. One such use case is for (un)pickling recursive closures (see heroic workaround here: https://github.com/cloudpipe/cloudpickle/pull/90/files#diff-d2a3618afedd4e124c532151eedbae09R74 ). Other use cases may include tinkering around and general education value.
There also doesn't seem to be, AFAICS, any counter-indication to being able to do so. It's already possible in C using PyCell_Set(), which is a public API. It just lacks an API in Python land. For example
cell_contentscould become a read/write property...lisroach commentedon May 27, 2017
This idea makes sense to me. I'll see if I can put together the code for it.
rhettinger commentedon May 27, 2017
+1 I don't see any reason this can't be writeable.
ncoghlan commentedon May 27, 2017
+1 from me, as this also has potential value in testing use cases and in dealing with some of the consequences of the zero-arg super() design (i.e. it will make the injected __class__ cell writable)
terryjreedy commentedon Jun 2, 2017
A possible counter-indication would be if setting a cell could cause a crash, as opposed to a mysterious exception. Since 3.x already allows writing any object to a cell from python code,
def outer():
cell = None
def inner(ob):
nonlocal cell
cell = ob # rebinds <hidden-cell>.cell_contents
return inner
making "cell.cell_contents = 'something'" legal should not enable more crashes.
I think that "function.__closure__[i].cell_contents = object" is perhaps not a good idea for production code, but I think it falls within the realm of 'consenting adults code' for the other uses suggested above. How should it be documented?
pitrou commentedon Jun 2, 2017
There's already a documentation change to the PR, you can leave your comments there.
serhiy-storchaka commentedon Jun 8, 2017
New changeset 64505a1 by Serhiy Storchaka (Lisa Roach) in branch 'master':
bpo-30486: Allow setting cell value (bpo-1840)
64505a1
serhiy-storchaka commentedon Jun 8, 2017
Thank you for your contribution Lisa!
lisroach commentedon Jun 9, 2017
Thank you for you guidance, Serhiy!
On Thu, Jun 8, 2017 at 4:45 AM, Serhiy Storchaka <report@bugs.python.org>
wrote:
vstinner commentedon Sep 20, 2017
New changeset 0ad05c3 by Victor Stinner in branch 'master':
bpo-30486: Make cell_set_contents() symbol private (bpo-3668)
0ad05c3