What fix should you pick for your “latest-comment” n+1 queries problem?

One example of this problem is trying to get the latest comment on a list of posts, but there are others, like the last review in a list of products, or the cheapest price, or the latest payment for each costumer, etc…

I have already shared some fixes to help you solve the problem. and some benchmarks for this fixes….

This is the code that I use to run those benchmarks. You can use it to play with the code and seed values, to make it more specific to your current problem.

Benchmark structure

The benchmark will include the memory and the “iterations per second” benchmarks.

With this “reports"…

• sorted_comments - Default order for the "has many” association
• latest_comment - A “has one” association for the latest comment
• feed looping - Looping through the latest comment for each post
• cached feed - A version of the looping where we cache feed in memory
• russian_doll - Russian doll-caching (with a simpler cache mechanism)
• cached_comment - Caching the latest comment in a column

The link

You can see the code on github at bhserna/latest_comment_n-1_fixes_benchmark

Related articles

• Styler and tachyons examples
• Pick a safe previous date to avoid race conditions saving computed values
• Extracting domain knowledge from a query method
• Examples to learn the difference between preload, includes or eager_load