Redmine

This patch optimizes Issue.load_visible_total_spent_hours for leaf issues. For a leaf issue, spent_hours and total_spent_hours are always the same because there are no descendants to include in the total. When visible spent_hours has already been preloaded for a leaf issue, the method reuses that value for total_spent_hours instead of running an expensive self-join aggregate query.

This improves performance in code paths that load both spent_hours and total_spent_hours, such as IssuesController#show and IssuesController#index views. In my development environment that uses MySQL, repeated requests to curl http://localhost:3000/issues/1.json with AnonymousUser reduced the query count from 22 to 20, and reduced average ActiveRecord time from 7.32 ms to 6.86 ms, which is about a 6.3% improvement.

Before:

Completed 200 OK in 15ms (Views: 3.0ms | ActiveRecord: 7.0ms (22 queries, 1 cached) | GC: 0.0ms)
Completed 200 OK in 30ms (Views: 3.4ms | ActiveRecord: 7.4ms (22 queries, 1 cached) | GC: 14.0ms)
Completed 200 OK in 17ms (Views: 3.6ms | ActiveRecord: 7.9ms (22 queries, 1 cached) | GC: 0.6ms)
Completed 200 OK in 16ms (Views: 3.1ms | ActiveRecord: 7.4ms (22 queries, 1 cached) | GC: 0.3ms)
Completed 200 OK in 16ms (Views: 3.5ms | ActiveRecord: 6.9ms (22 queries, 1 cached) | GC: 0.3ms)

After:

Completed 200 OK in 16ms (Views: 3.4ms | ActiveRecord: 8.0ms (20 queries, 0 cached) | GC: 0.0ms)
Completed 200 OK in 13ms (Views: 3.7ms | ActiveRecord: 6.4ms (20 queries, 0 cached) | GC: 0.0ms)
Completed 200 OK in 14ms (Views: 3.3ms | ActiveRecord: 6.4ms (20 queries, 0 cached) | GC: 0.0ms)
Completed 200 OK in 14ms (Views: 3.2ms | ActiveRecord: 6.1ms (20 queries, 0 cached) | GC: 0.0ms)
Completed 200 OK in 15ms (Views: 3.7ms | ActiveRecord: 7.4ms (20 queries, 0 cached) | GC: 0.0ms)