Remove "descendant_count" to optimize large taxonomies

[bradenmacdonald]

Hmm, I guess when testing my previous PR #511 I didn't test it on a sufficiently large taxonomy.

Once I tested with Lightcast Open Skills Taxonomy.csv (4,268 skill tags in a 3-level hierarchy), I realized that it was slower for very large taxonomies, going from ~150ms to ~14s :/

The reason is due to computing the descendant_count field accurately.

Technical Details

I thought that this `lineage__startswith` query would be performant, because we have an index on the `lineage` column and it's a `startswith` query:

        # Count all descendants at any depth using depth + lineage prefix.
        # depth__gt correctly excludes self; lineage prefix matches all descendants.
        descendants_sq = (
            self.tag_set.filter(depth__gt=models.OuterRef("depth"), lineage__startswith=models.OuterRef("lineage"))
            .order_by()
            .annotate(count=models.Func(F("id"), function="Count"))
        )
        qs = qs.annotate(descendant_count=models.Subquery(descendants_sq.values("count")))  # type: ignore[no-redef]

But, SQL doesn't natively have a "starts with" operator, and this gets converted to a LIKE "...%" operator. The problem is that because this is a subquery, our LIKE "..." expression is using a variable, the OuterRef("lineage") from the outer query, and so we run into a fundamental MySQL optimizer limitation:

for a correlated subquery, MySQL plans the inner query once before executing it. Since CONCAT(outer_ref, '%') is a runtime value, MySQL can't compute range bounds at plan time and won't apply a range scan — even with the perfect index.

This makes the query O(n²).

Instead of finding a way to optimize descendant_count in this query (I don't think there is a good way, other than the hard-coded approach we had before), I think it's better to just remove descendant_count completely. It's expensive to compute in the database, but it's almost trivial to compute afterward in python once you've run the same query. Also, we aren't using it at all (I checked frontend-platform and the Authoring MFE). So I don't think we should compute it or return it, when we don't have a use case. The existing child_count is far more important for things like pagination purposes, and it works fine.

Q: Why don't I just compute descendant_count in python within the get_filtered_tags_deep code?

A: Because doing so requires evaluating the queryset, and I want this low-level API to return a queryset that can then be paginated or filtered further. That way is much more performant then pre-evaluating the entire query for the whole taxonomy.

Testing Instrucions

Import the linked taxonomy above and test its performance before and after this change.

Private ref: MNG-4914

[openedx-webhooks]

Thanks for the pull request, @bradenmacdonald!

This repository is currently maintained by @axim-engineering.

Once you've gone through the following steps feel free to tag them in a comment and let them know that your changes are ready for engineering review.

🔘 Get product approval

If you haven't already, check this list to see if your contribution needs to go through the product review process.

• If it does, you'll need to submit a product proposal for your contribution, and have it reviewed by the Product Working Group.
  • This process (including the steps you'll need to take) is documented here.
• If it doesn't, simply proceed with the next step.

🔘 Provide context

To help your reviewers and other members of the community understand the purpose and larger context of your changes, feel free to add as much of the following information to the PR description as you can:

• Dependencies

  This PR must be merged before / after / at the same time as ...

• Blockers

  This PR is waiting for OEP-1234 to be accepted.

• Timeline information

  This PR must be merged by XX date because ...

• Partner information

  This is for a course on edx.org.

• Supporting documentation
• Relevant Open edX discussion forum threads

🔘 Get a green build

If one or more checks are failing, continue working on your changes until this is no longer the case and your build turns green.

Details

---

Where can I find more information?

If you'd like to get more details on all aspects of the review process for open source pull requests (OSPRs), check out the following resources:

• Overview of Review Process for Community Contributions
• Pull Request Status Guide
• Making changes to your pull request

When can I expect my changes to be merged?

Our goal is to get community contributions seen and reviewed as efficiently as possible.

However, the amount of time that it takes to review and merge a PR can vary significantly based on factors such as:

• The size and impact of the changes that it introduces
• The need for product review
• Maintenance status of the parent repository

💡 As a result it may take up to several weeks or months to complete a review and merge your PR.