Reordering nodes can cause #ancestor to give results out of order

[benhutton]

Here's some sample output.

This is the order that my parent ids are stored in the array:

irb(main):018:0> page.parent_ids
=> ["515c547c392d4ecdf6000004", "515c5732392d4ecdf600003a", "51d1f63fd268d9289600037c", "515c5740392d4ecdf60000f1"]

This is the order that they are returned by #ancestors:

irb(main):004:0> page.ancestors.map(&:id)
=> ["515c547c392d4ecdf6000004", "515c5732392d4ecdf600003a", "515c5740392d4ecdf60000f1", "51d1f63fd268d9289600037c"]

You'll notice the last 2 are flipped.

It seems that when no sort order is specified, MongoDB returns documents based on the order that they are stored in the disk: http://docs.mongodb.org/manual/reference/glossary/#term-natural-order

It seems to me that the least computationally expensive way to do this is to add a depth field, which may come in handy other places as well. I'm not very experienced with mongodb, so if anyone has a better suggestion I'm very open to it.

[benedikt]

Hi Ben,

thanks for pointing this out and your pull request! I'll look into this more in depth over the weekend. :-)

Cheers,

Benedikt

[benedikt]

Hi Ben,

thought about this over the weekend. What about ordering by parent_ids? As every ancestor gets one more parent_id in the parent_ids array this should output all parent ids in the correct order, right?

Bests,

Benedikt

[benhutton]

This was my first thought, and would be the ideal way to implement it. parent_ids array order is the canonical ordering as far as I'm aware.

But like I said, my mongodb skills are too limited to know how to actually do this in a mongodb call, and it seemed like it was important to keep the call to #ancestors returning a mongoid criterion.

If you can point me in the right direction, I can make the changes.

[hems]

👍
nice stuff guys

[benhutton]

Awesome, thanks @benedikt !

One question: do you also need to apply your fix to the Ordering module here: https://github.com/benedikt/mongoid-tree/blob/master/lib/mongoid/tree/ordering.rb#L49-L51 ?

[benedikt]

Uh oh… no… I missed that :) Thanks for pointing out! 👍