Give point-in-time copies their own stackable and rescue-handler stores

[ericproulx]

Summary

InheritableSetting#point_in_time_copy copied a scope's stackable store and its rescue-handler maps shallowly, so the nested Arrays and Hashes stayed shared with the source. A registration made after an endpoint was defined therefore still reached that endpoint — but only when the key already held a registration at the time the copy was taken, since otherwise the writer allocated a fresh store on the source alone.

The outcome depended on something the API never expressed:

use Middleware1
get('/x') { }
use Middleware2          # applied to GET /x

get('/x') { }
use Middleware2          # did NOT apply to GET /x

rescue_from(ArgumentError) { }
get('/x') { raise Boom }
rescue_from(Boom) { }    # rescued GET /x

get('/x') { raise Boom }
rescue_from(Boom) { }    # did NOT rescue GET /x

	before	after
use MW1 · route · use MW2	MW1, MW2	MW1
route · use MW2	—	—
rescue_from A · route · rescue_from B	B rescues the route	raises
route · rescue_from B	raises	raises

Two APIs stating the same thing, behaving differently. Not a deliberate "late registration" feature either — helpers defined after an endpoint already did not leak, because they resolve down a different path.

Approach

Dup the nested stores as well as the Hash holding them. A copy is a point in time: what the source registers afterwards must not reach it.

Inheritance is untouched — it resolves by walking #parent, so a scope still sees values an enclosing scope gains later. That is what the existing decouples namespace stackable values spec actually exercised: its value lives on the parent, so nothing was ever shared and it passed with or without this change. The new specs register on the scope itself before taking the copy, which is the shape that leaked.

Backward compatibility

UPGRADING entry added. An API that declares use / helpers / before / rescue_from below its routes and relies on it applying to them will now see it silently not run. That arrangement only ever worked when an earlier registration for the same key happened to seed the store, so it was never dependable; the fix is to move the registration above the routes it should cover, which is where the documentation has always placed it.

Perf

Definition-time only. 300 endpoints under scopes carrying 8 middleware, 4 helpers and 6 filters:

	allocations	boot
before	463,062	0.127s
after	464,862 (+0.4%)	0.128s

Test plan

• 5 new unit examples in inheritable_setting_spec.rb (stackable and rescue-handler leakage in both directions, sibling independence, base-only handlers) and 4 integration examples in api_spec.rb covering use and rescue_from in both the seeded and unseeded arrangements; verified the behavioural ones fail without the lib/ change.
• Full RSpec suite passes locally (2552 examples, 0 failures).
• RuboCop clean.
• CI green.

🤖 Generated with Claude Code

[github-actions]

Danger Report

No issues found.

View run