bpo-30256: pass all BaseProxy arguments through AutoProxy

[uSpike]

The AutoProxy method did not accept the "manager_owned" keyword argument which broke when nested AutoProxy objects were created.

https://bugs.python.org/issue30256

[the-knights-who-say-ni]

Hello, and thanks for your contribution!

I'm a bot set up to make sure that the project can legally accept your contribution by verifying you have signed the PSF contributor agreement (CLA).

Unfortunately we couldn't find an account corresponding to your GitHub username on bugs.python.org (b.p.o) to verify you have signed the CLA (this might be simply due to a missing "GitHub Name" entry in your b.p.o account settings). This is necessary for legal reasons before we can look at your contribution. Please follow the steps outlined in the CPython devguide to rectify this issue.

Thanks again to your contribution and we look forward to looking at it!

[uSpike]

I signed the CLA (over a month ago) and my bpo account reflects that, however @the-knights-who-say-ni has not updated yet.

[Phhere]

When will this get merged ?

[finefoot]

I stumbled upon this problem, too: https://stackoverflow.com/questions/56716470/python-multiprocessing-nested-shared-objects-doesnt-work-with-queue

Right now, Python docs state

Shared objects are capable of being nested. For example, a shared container object such as a shared list can contain other shared objects which will all be managed and synchronized by the SyncManager.

These changes have been introduced with https://hg.python.org/cpython/rev/39e7307f9aee along with some tests. However, the tests only use nested ListProxy and DictProxy.

So maybe this PR should add tests for AutoProxy[Queue], too?

---

Update: Some tests have been added with #16341

[Phhere]

@uSpike could you add an entry to Misc/News.d/next for this issue? maybe it get merged when CI is fine

[carlbordum]

Hey I just want to say that this patch fixed my issue.

[brandtbucher]

@pitrou, care to review? It looks like there's at least one PR waiting on this.

I'm not too familiar with multiprocessing, but the change looks good. The only thing that I think should be added is a NEWS entry (#16341 contains a regression test from a first-timer, but it can't be merged yet).

[uSpike]

@brandtbucher I've added a NEWS entry.

[brandtbucher]

Awesome, thanks for this @uSpike. And welcome to CPython!

[Phhere]

When will it be merged ?

I think @applio it not active any more on this project.
@pitrou can you review it ?

[jimenez-alex]

@pitrou | ( @Phhere ) any ETA for this simple change to be reviewed?

[tstordyallison]

Any update on this one? Feels like a very easy fix :)

@applio around for a quick approval?

[Phhere]

Maybe one of these can help:
@pitrou
@vstinner

[uSpike]

Gentle bump? This PR has been open for 2.5 years, it's a trivial change, and I'd love to be a contributor! I really don't want to miss the window for 3.9.

[nanjekyejoannah]

Update : I was wondering it a test exists and I saw the test here : #16341 sorry.

[applio]

When testing this fix, this post on StackOverflow suggests they experienced a seg fault: https://stackoverflow.com/questions/56716470/multiprocessing-manager-nested-shared-objects-doesnt-work-with-queue

Has this been investigated and resolved?

[Phhere]

I was not able to reproduce any seg faults with this patch

[StefanD986]

I tried this patch and I did not get any segfaults.

Would be nice if this could get fixed into python 3.9.

My system:
Python 3.7.9 (tags/v3.7.9:13c94747c7, Aug 17 2020, 18:58:18) [MSC v.1900 64 bit (AMD64)] on win32

[uSpike]

I'm also unable to reproduce any seg faults in 3.7.5 on Ubuntu 20.04. I may be wrong, but I think the stackoverflow message that @applio is referencing is not suggesting that this patch causes a seg fault. The code he says "works" is the same as this patch.

[uSpike]

I'm closing this in favor of #16341