Add an option (possibly the default?) for DeepCollectionEquality to handle self-referential structures

[DartBot]

<img src="https://avatars.githubusercontent.com/u/188?v=3" align="left" width="96" height="96"hspace="10"> Issue by nex3 Originally opened as dart-lang/sdk#18729

---

Sometimes it's useful to be able to compare nested object structures that can contain self-referential loops. Currently DeepCollectionEquality causes a stack overflow when passed such structures, but it would be nice if it could handle them. This could be done either as a configuration option or as the default if the prospect of storing back-references isn't too worrisome.

Currently the YAML package has its own custom implementation of deep equality, but it would be nice if it could use the standard implementation instead.

[DartBot]

<img src="https://avatars.githubusercontent.com/u/4865287?v=3" align="left" width="48" height="48"hspace="10"> Comment by lrhn

---

That should be doable. For each operand, remember each object it is currently being compared to, and if you see that combination again, return true. It's not fool-proof, but it will catch most simple self-referential structures.

[DartBot]

<img src="https://avatars.githubusercontent.com/u/188?v=3" align="left" width="48" height="48"hspace="10"> Comment by nex3

---

I believe YAML's deepEquals operation works like that, but it remembers the order of the objects it's being compared to as well.

[DartBot]

<img src="https://avatars.githubusercontent.com/u/17034?v=3" align="left" width="48" height="48"hspace="10"> Comment by kevmoo

---

Aligning w/ the bits offered from matcher would be nice, too.

Fewer implementations FTW

[DartBot]

<img src="https://avatars.githubusercontent.com/u/188?v=3" align="left" width="48" height="48"hspace="10"> Comment by nex3

---

Matcher's use case is a little different, because it needs to track a bunch of extra information to provide useful messaging when the equality fails.

[matanlurey]

This has been open since 2015 without action so I am closing as stale.

[nex3]

@matanlurey I appreciate your zeal for cleaning up the issue tracker, but it's useful for me as a maintainer of this package to track features that would be useful to users, even if I don't often get a chance to work on those features. As long as these features would still be useful, I'd like to keep the tracking issues open.

[lrhn]

I don't want to make the existing deep equality check for cycles. That's a lot of work (and memory for caching) which has no effect in most cases.

We can add another equivalence which does handle cycles, if we want to.