Firebreak.WhatIf (Firebreak v0.1.0)

Copy Markdown View Source

Topology refactor simulation: mutate the supervision structure, re-run the real analysis on the hypothetical, and diff the synchronous cross-tree crossings — so you can answer "does this move actually fix the coupling, or just relocate it?" before touching code.

The mutation is applied to the parsed modules and fed back through Firebreak.analyze_modules/1, so the result uses the same forest/coupling/check logic as a normal run — nothing is re-implemented or approximated. The diff is computed over the model IR's synchronous crossings.

a = Firebreak.analyze("my_app")
Firebreak.WhatIf.move(a, MyApp.Cache, MyApp.Worker.Supervisor)
#=> %{resolved: [{MyApp.Worker, MyApp.Cache}], introduced: [], before: 1, after: 0}

move/3 relocates a child to a new supervisor (the common fix for a cross_tree_coupling finding — colocate the caller and target). set_strategy/3 changes a supervisor's restart strategy; it's expected to leave the crossing set unchanged (strategy doesn't fix coupling), which is itself a useful thing to confirm.

Summary

Types

diff()

Functions

move(a, child, new_parent)

Simulate relocating child to be a child of new_parent; diff the crossings.

set_strategy(a, sup, strategy)

Simulate changing sup's restart strategy; diff the crossings.

Types

diff()

@type diff() :: %{
  resolved: [{module(), module()}],
  introduced: [{module(), module()}],
  before: non_neg_integer(),
  after: non_neg_integer()
}

Functions

move(a, child, new_parent)

@spec move(Firebreak.Analysis.t(), module(), module()) :: diff()

Simulate relocating child to be a child of new_parent; diff the crossings.

set_strategy(a, sup, strategy)

@spec set_strategy(Firebreak.Analysis.t(), module(), atom()) :: diff()

Simulate changing sup's restart strategy; diff the crossings.