Implement Weak::{strong_count, weak_count} #56696

jonas-schievink · 2018-12-10T23:57:47Z

The counters are also useful on Weak, not just on strong references (Rc or Arc).

In situations where there are still strong references around, you can also get these counts by temporarily upgrading and adjusting the values accordingly. Using the methods introduced here is simpler to do, less error-prone (since you can't forget to adjust the counts), can also be used when no strong references are around anymore, and might be more efficient due to not having to temporarily create an Rc.

This is mainly useful in assertions or tests of complex data structures. Data structures might have internal invariants that make them the sole owner of a Weak pointer, and an assertion on the weak count could be used to ensure that this indeed happens as expected. Due to the presence of Weak::upgrade, the strong_count becomes less useful, but it still seems worthwhile to mirror the API of Rc.

TODO:

Tracking issue - Tracking issue for Weak::{weak_count, strong_count} #57977

Closes #50158

jonas-schievink · 2018-12-10T23:59:02Z

src/liballoc/sync.rs

+    // Due to the implicit weak pointer added when any strong pointers are
+    // around, we cannot implement `weak_count` correctly since it necessarily
+    // requires accessing the strong count and weak count in an unsynchronized
+    // fashion.


I've added a strong_count method to sync::Weak as well, but here the tradeoff is less clear since (I think) weak_count can not be implemented correctly, so the API must stay asymmetric like this (or not offer strong_count at all, since it isn't very useful)

Dylan-DPC-zz · 2019-01-07T17:54:15Z

ping from triage, can anyone from @rust-lang/libs review this?

alexcrichton · 2019-01-07T18:44:51Z

Having a discrepancy in weak_count is sort of a bummer, would it be possible to implement, even a racy version, on Arc? Otherwise shouldn't weak_count on a Weak always return >= 1?

Centril · 2019-01-25T16:55:10Z

r? @alexcrichton

alexcrichton · 2019-01-25T19:50:57Z

Oops sorry for the delay, @jonas-schievink feel free to ping the PR when it's updated as notifications aren't always sent out.

I wonder if weak_count here should return Option<usize> where None is returned if all the strong references have been dropped and as a result the weak count may not be known?

jonas-schievink · 2019-01-25T20:20:16Z

Weak::weak_count only returns 0 if the Weak was created via Weak::new(), strong references don't matter. I do agree that returning None in that case might be better since having a method on Weak tell you "there are no Weaks" is a bit contradictory.

Should the same be done for strong_count? That one is less contradictory, but "There are no strong pointers pointing to the value" (0) is still more wrong than returning "There is no value any strong pointer could point to" (None). Having strong_count still return usize would also make for an odd difference between two otherwise very similar methods.

alexcrichton · 2019-01-28T22:33:30Z

Yeah I agree that Weak::new is the only problematic case to handle here. I think for strong_count though it's fine to infer zero because Weak::new does indeed imply there are only ever zero strong pointers!

alexcrichton · 2019-01-28T22:34:11Z

Er to add some more, I'm not too worried about the inconsistency of return values, these are pretty niche methods so having them be absolutely consistent isn't so important

alexcrichton · 2019-01-31T16:43:56Z

@bors: r+

Ok looks great to me to add, thanks @jonas-schievink!

bors · 2019-01-31T16:43:57Z

📌 Commit 0d314f0 has been approved by alexcrichton

bors · 2019-01-31T19:20:22Z

⌛ Testing commit 0d314f0 with merge f29b4fb...

Implement Weak::{strong_count, weak_count} The counters are also useful on `Weak`, not just on strong references (`Rc` or `Arc`). In situations where there are still strong references around, you can also get these counts by temporarily upgrading and adjusting the values accordingly. Using the methods introduced here is simpler to do, less error-prone (since you can't forget to adjust the counts), can also be used when no strong references are around anymore, and might be more efficient due to not having to temporarily create an `Rc`. This is mainly useful in assertions or tests of complex data structures. Data structures might have internal invariants that make them the sole owner of a `Weak` pointer, and an assertion on the weak count could be used to ensure that this indeed happens as expected. Due to the presence of `Weak::upgrade`, the `strong_count` becomes less useful, but it still seems worthwhile to mirror the API of `Rc`. TODO: * [X] Tracking issue - #57977 Closes #50158

bors · 2019-01-31T22:13:55Z

☀️ Test successful - checks-travis, status-appveyor
Approved by: alexcrichton
Pushing f29b4fb to master...

@bdonlan

Stabilize `std::{rc,sync}::Weak::{weak_count, strong_count}` * Original PR: rust-lang#56696 * Tracking issue: rust-lang#57977 Closes: rust-lang#57977 Supporting comments: > Although these were added for testing, it is occasionally useful to have a way to probe optimistically for whether a weak pointer has become dangling, without actually taking the overhead of manipulating atomics. Are there any plans to stabilize this? _Originally posted by @bdonlan in rust-lang#57977 (comment) > Having this stabilized would help. Currently, the only way to check if a weak pointer has become dangling is to call `upgrade`, which is by far expensive. _Originally posted by @glebpom in rust-lang#57977 (comment) Not sure if stabilizing these warrants a full RFC, so throwing this out here as a start for now. Note: per CONTRIBUTING.md, I ran the tidy checks, but they seem to be failing on unchanged files (primarily in `src/stdsimd`).

@bdonlan

Stabilize `std::{rc,sync}::Weak::{weak_count, strong_count}` * Original PR: rust-lang#56696 * Tracking issue: rust-lang#57977 Closes: rust-lang#57977 Supporting comments: > Although these were added for testing, it is occasionally useful to have a way to probe optimistically for whether a weak pointer has become dangling, without actually taking the overhead of manipulating atomics. Are there any plans to stabilize this? _Originally posted by @bdonlan in rust-lang#57977 (comment) > Having this stabilized would help. Currently, the only way to check if a weak pointer has become dangling is to call `upgrade`, which is by far expensive. _Originally posted by @glebpom in rust-lang#57977 (comment) Not sure if stabilizing these warrants a full RFC, so throwing this out here as a start for now. Note: per CONTRIBUTING.md, I ran the tidy checks, but they seem to be failing on unchanged files (primarily in `src/stdsimd`).

@bdonlan

Stabilize `std::{rc,sync}::Weak::{weak_count, strong_count}` * Original PR: rust-lang#56696 * Tracking issue: rust-lang#57977 Closes: rust-lang#57977 Supporting comments: > Although these were added for testing, it is occasionally useful to have a way to probe optimistically for whether a weak pointer has become dangling, without actually taking the overhead of manipulating atomics. Are there any plans to stabilize this? _Originally posted by @bdonlan in rust-lang#57977 (comment) > Having this stabilized would help. Currently, the only way to check if a weak pointer has become dangling is to call `upgrade`, which is by far expensive. _Originally posted by @glebpom in rust-lang#57977 (comment) Not sure if stabilizing these warrants a full RFC, so throwing this out here as a start for now. Note: per CONTRIBUTING.md, I ran the tidy checks, but they seem to be failing on unchanged files (primarily in `src/stdsimd`).

rust-highfive assigned KodrAus Dec 10, 2018

This comment has been minimized.

Sign in to view

rust-highfive added the S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. label Dec 10, 2018

jonas-schievink commented Dec 10, 2018

View reviewed changes

rust-highfive assigned alexcrichton and unassigned KodrAus Jan 25, 2019

jonas-schievink added 3 commits January 29, 2019 21:07

Implement Weak::{strong_count, weak_count}

2fe3b3b

Implement a slightly racy sync::Weak::weak_count

7e0edb3

Make weak_count return an Option<usize>

b664341

jonas-schievink mentioned this pull request Jan 29, 2019

Tracking issue for Weak::{weak_count, strong_count} #57977

Closed

Add tracking issue to unstable attribute

0d314f0

bors added S-waiting-on-bors Status: Waiting on bors to run and complete tests. Bors will change the label on completion. and removed S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. labels Jan 31, 2019

bors merged commit 0d314f0 into rust-lang:master Jan 31, 2019

bdonlan mentioned this pull request Oct 24, 2019

Stabilize std::{rc,sync}::Weak::{weak_count, strong_count} #65778

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Implement Weak::{strong_count, weak_count} #56696

Implement Weak::{strong_count, weak_count} #56696

jonas-schievink commented Dec 10, 2018 •

edited

Loading

This comment has been minimized.

jonas-schievink Dec 10, 2018 •

edited

Loading

Dylan-DPC-zz commented Jan 7, 2019

alexcrichton commented Jan 7, 2019

Centril commented Jan 25, 2019

alexcrichton commented Jan 25, 2019

jonas-schievink commented Jan 25, 2019

alexcrichton commented Jan 28, 2019

alexcrichton commented Jan 28, 2019

alexcrichton commented Jan 31, 2019

bors commented Jan 31, 2019

bors commented Jan 31, 2019

bors commented Jan 31, 2019

Implement Weak::{strong_count, weak_count} #56696

Implement Weak::{strong_count, weak_count} #56696

Conversation

jonas-schievink commented Dec 10, 2018 • edited Loading

This comment has been minimized.

jonas-schievink Dec 10, 2018 • edited Loading

Choose a reason for hiding this comment

Dylan-DPC-zz commented Jan 7, 2019

alexcrichton commented Jan 7, 2019

Centril commented Jan 25, 2019

alexcrichton commented Jan 25, 2019

jonas-schievink commented Jan 25, 2019

alexcrichton commented Jan 28, 2019

alexcrichton commented Jan 28, 2019

alexcrichton commented Jan 31, 2019

bors commented Jan 31, 2019

bors commented Jan 31, 2019

bors commented Jan 31, 2019

jonas-schievink commented Dec 10, 2018 •

edited

Loading

jonas-schievink Dec 10, 2018 •

edited

Loading