Borda
diff --git a/‎.github/CONTRIBUTING.md‎
Lines changed: 11 additions & 1 deletion b/‎.github/CONTRIBUTING.md‎
Lines changed: 11 additions & 1 deletion
diff --git a/‎AGENTS.md‎
Lines changed: 2 additions & 0 deletions b/‎AGENTS.md‎
Lines changed: 2 additions & 0 deletions
diff --git a/‎CHANGELOG.md‎
Lines changed: 2 additions & 0 deletions b/‎CHANGELOG.md‎
Lines changed: 2 additions & 0 deletions
diff --git a/‎README.md‎
Lines changed: 1 addition & 0 deletions b/‎README.md‎
Lines changed: 1 addition & 0 deletions
diff --git a/‎docs/guide/use-cases.md‎
Lines changed: 172 additions & 19 deletions b/‎docs/guide/use-cases.md‎
Lines changed: 172 additions & 19 deletions
@@ -374,7 +374,7 @@ Tests live in `tests/` and follow a **three-layer separation**:
 > [!IMPORTANT]
 > **README examples must be runnable.** Every Python code block in `README.md` is extracted by `phmdoctest` and executed as a test. Follow these rules when writing README examples:
 >
-> - Use `print()` for values you want to verify, paired with a `<details><summary>Output: ...</summary>` block immediately after the code block.
+> - Use `print()` for values you want to verify, paired with a `<details><summary>Output: <code>expression</code></summary>` block immediately after the code block. The `<summary>` label shows the **expression** being evaluated (e.g. `cfg.timeout`), not the `print()` wrapper.
 > - Only import and use `pytest.raises` when an example intentionally raises an exception — this prevents the extracted test from crashing. Do **not** use `pytest.warns`; deprecation warnings are emitted to stderr and do not cause test failures.
 > - Do **not** use bare `assert` statements — they crash the test with an unhelpful `AssertionError` if the value changes.
 > - Regenerate `test_readme.py` after any README change: `phmdoctest README.md --outfile tests/integration/test_readme.py`
@@ -486,6 +486,16 @@ def decorated_sum_warn_only(a: int, b: int = 5) -> int:
 - **One behavior per test** — each test method should verify one specific aspect.
 - **Prefer parametrization for repetitive shapes** — when the setup/assertion flow is the same and only inputs/expected outputs differ, use `pytest.mark.parametrize(...)` to reduce duplication while keeping one behavioral intent per case.
 - **Assertions on warnings:** Use `pytest.warns(FutureWarning|DeprecationWarning)` to verify deprecation warnings are emitted correctly.
+- **Scenario description in docstrings** — every non-trivial test method must include a prose paragraph after the one-line summary that describes the real-world situation being tested. A one-line summary alone is not sufficient for complex tests.
+
+```python
+def test_warns_on_read(self) -> None:
+    """FutureWarning fires on property read access.
+
+    A user accesses a deprecated property on an existing object and
+    expects a FutureWarning with the original value still returned.
+    """
+```
 
 **For bug fixes:**
 
 
@@ -75,6 +75,7 @@ Write a clear explanation linking to both sources, then let maintainers decide o
 - Help onboard new contributors
 - Ensure deprecation examples are documented in both README and `docs/guide/use-cases.md`
 - Keep `docs/troubleshooting.md` and its FAQPage JSON-LD in `docs/overrides/main.html` in sync
+- Every `docs/**/*.md` code block containing `print()` must be followed by a `<details><summary>Output: <code>…</code></summary>` block with verified output (see [Docs examples](.github/CONTRIBUTING.md#test-organization))
 
 **Guidelines**:
 
@@ -121,6 +122,7 @@ Write a clear explanation linking to both sources, then let maintainers decide o
 
 - **Three-layer separation**: targets in `collection_targets.py`, deprecated wrappers in `collection_deprecate.py`, test logic in `test_*.py`
 - **Do not** define targets or `@deprecated` wrappers directly in test files
+- **Scenario description required** — every non-trivial test method docstring must include a prose paragraph describing the real-world situation being tested; a one-line summary alone is not sufficient (see [Test Requirements](.github/CONTRIBUTING.md#-tests-and-quality-assurance))
 - See [Test Organization](.github/CONTRIBUTING.md#test-organization) for details
 
 ### Documentation Site
 
@@ -4,6 +4,8 @@
 
 ### Added
 
+- **`@deprecated @property` now wraps `fset` and `fdel` with `FutureWarning`.** Applying `@deprecated` on the outside of `@property` (outer order, or explicit `deprecated(...)(property(fget, fset, fdel))`) now wraps all three accessors. Previously, only `fget` emitted a warning; `fset` and `fdel` were silently passed through. Consumers running `filterwarnings=error::FutureWarning` that wrote to or deleted a deprecated property will now see `FutureWarning` errors — use inner-order (`@property @deprecated`) or decorate only `fget` directly if you want a silent setter/deleter. Chain-style rebinding via `@value.setter` / `@value.deleter` is fully supported through the new `_DeprecatedProperty` subclass. ([#190](https://github.com/Borda/pyDeprecate/pull/190))
+
 ### Changed
 
 ### Deprecated
 
@@ -71,6 +71,7 @@ Another good aspect is not overwhelming users with too many warnings, so per fun
 - 🚫 The deprecated function body is never executed when using `target`
 - ⚡ Minimal runtime overhead with zero dependencies (Python standard library only)
 - 🛠️ Supports deprecating callables: functions, methods, class constructors, **generator functions** (`def gen(): yield` — warning fires eagerly at call time, before iteration begins), and **`async def` coroutine functions** (wrapper is `async def`, `inspect.iscoroutinefunction(wrapper)` returns `True`; warning fires when the coroutine is awaited)
+- 🏠 Supports deprecating `@property` — all three accessors (`fget`, `fset`, `fdel`) fire `FutureWarning`; chain-style `.setter()` / `.deleter()` rebinding preserved via `_DeprecatedProperty` subclass. Also supports `@cached_property` (getter-only — no `fset`/`fdel`).
 - 📦 Supports deprecating classes, Enums, dataclasses, and module-level constants/objects via transparent proxies (`deprecated_class` for types; `deprecated_instance` for objects, with optional read-only enforcement — blocks standard collection mutators: `append`, `pop`, `update`, `clear`, etc.; custom mutator names bypass this guard)
 - 📝 Optionally, docstrings can be updated automatically in RST/Sphinx or MkDocs/Markdown format (auto-detected); ships bundled Griffe and Sphinx doc-engine extensions
 - 🔍 Preserves original function signature, annotations and metadata for introspection
 
@@ -976,7 +976,7 @@ Both decorator orders produce `classmethod(deprecated_wrapper)` or `staticmethod
 
 ## Properties and cached properties
 
-`@deprecated` works with `@property` and `@cached_property` in either decorator order — the deprecation warning fires correctly at access time regardless of which order the decorators were applied.
+`@deprecated` works with `@property` and `@cached_property`. The decorator only adds a `FutureWarning` at access time — it does **not** forward reads or writes to another property. For a getter-only property, either decorator order is valid. To add a warning to all three accessors (`fget`, `fset`, `fdel`) so that read, write, **and** delete each fire `FutureWarning`, place `@deprecated` on the **outside** (`@deprecated @property` order, or explicit `deprecated(...)(property(fget, fset, fdel))`). The inner-first order (`@property @deprecated`) only adds a warning to `fget` — apply `@deprecated` to setter and deleter separately if you also need them to warn.
 
 ```python
 from functools import cached_property
@@ -985,48 +985,201 @@ from deprecate import deprecated
 
 
 class Config:
-    # @deprecated inside @property — conventional order
     @property
     @deprecated(deprecated_in="1.0", remove_in="2.0")
     def timeout(self) -> int:
         return 30
 
-    # @deprecated outside @property — also works
-    @deprecated(deprecated_in="1.0", remove_in="2.0")
-    @property
-    def retries(self) -> int:
-        return 3
-
-    # @deprecated inside @cached_property — conventional order
     @cached_property
     @deprecated(deprecated_in="1.0", remove_in="2.0")
     def base_url(self) -> str:
         return "https://example.com"
 
-    # @deprecated outside @cached_property — also works
+
+print(Config().timeout)
+```
+
+<details>
+  <summary>Output: <code>Config().timeout</code></summary>
+
+```
+30
+```
+
+</details>
+
+The `FutureWarning` fires on **attribute access** (`obj.timeout`), not on a call. For `@cached_property`, the warning fires on **first access only** — subsequent accesses return the cached value without emitting another warning.
+
+!!! tip "Decorator order for getter-only properties"
+
+    Either `@property @deprecated` (inner) or `@deprecated @property` (outer) order works for getter-only properties. Inner order is conventional — the deprecated decorator is closer to the `def`. For properties with a setter or deleter, use outer order; see the next section.
+
+### Deprecating a property with a setter or deleter
+
+When the property being deprecated has a setter or deleter, all three accessors (`fget`, `fset`, `fdel`) are wrapped automatically — each fires a `FutureWarning`. Both the chain-style decorator pattern and the explicit construction pattern work:
+
+```python
+from deprecate import deprecated
+
+
+class Config:
+    def __init__(self) -> None:
+        self._timeout: int = 30
+
+    # Outer order required: @deprecated @property wraps fget, fset, and fdel
     @deprecated(deprecated_in="1.0", remove_in="2.0")
-    @cached_property
-    def legacy_url(self) -> str:
-        return "https://old.example.com"
+    @property
+    def timeout(self) -> int:
+        return self._timeout
+
+    @timeout.setter
+    def timeout(self, value: int) -> None:
+        self._timeout = value
 
+    @timeout.deleter
+    def timeout(self) -> None:
+        del self._timeout
 
-print(Config().legacy_url)
+
+cfg = Config()
+cfg.timeout = 10  # FutureWarning: write
+print(cfg.timeout)  # FutureWarning: read; prints 10
+del cfg.timeout  # FutureWarning: delete
 ```
 
 <details>
-  <summary>Output: <code>Config().legacy_url</code></summary>
+  <summary>Output: <code>cfg.timeout</code></summary>
 
 ```
-https://old.example.com
+10
 ```
 
 </details>
 
-The `FutureWarning` fires on **attribute access** (`obj.timeout`), not on a call. For `@cached_property`, the warning fires on **first access only** — subsequent accesses return the cached value without emitting another warning.
+`obj.timeout` fires `FutureWarning` on **read**, `obj.timeout = value` fires on **write**, and `del obj.timeout` fires on **delete**.
+
+!!! tip "Want only the getter to warn?"
+
+    If you want the setter or deleter to remain silent, apply `@deprecated` directly to `fget` using inner order (`@property @deprecated`) instead of wrapping the full `property` object.
+
+!!! tip "Testing each accessor independently"
+
+    Each accessor (`fget`, `fset`, `fdel`) has its own warning counter — assert read, write, and delete warnings in separate `pytest.warns` blocks, or use `num_warns=-1` to disable per-accessor deduplication.
+
+The explicit `property(fget, fset[, fdel])` construction also works:
+
+```python
+from deprecate import deprecated
+
+
+def _timeout_fget(self) -> int:
+    return self._timeout
+
+
+def _timeout_fset(self, value: int) -> None:
+    self._timeout = value
+
+
+def _timeout_fdel(self) -> None:
+    del self._timeout
+
+
+class Config:
+    def __init__(self) -> None:
+        self._timeout: int = 30
+
+    timeout = deprecated(deprecated_in="1.0", remove_in="2.0")(property(_timeout_fget, _timeout_fset, _timeout_fdel))
+```
+
+!!! note "Audit discoverability with explicit construction"
+
+    `find_deprecation_wrappers` discovers explicit-construction properties via the accessor that carries `__deprecated__` metadata. For setter-only properties (`property(None, fset)`), it discovers via `fset`; if `fget` is plain (not deprecated), it falls through to `fset` or `fdel`.
+
+### Deprecated property alias on a dataclass
+
+When a dataclass field is renamed, define a property with the old name that delegates to the new field in its accessor body. `@deprecated` adds a `FutureWarning` to each accessor — the delegation itself is plain Python in the method body, not something the library provides.
+
+**Read-only alias (warns on read only):**
+
+```python
+from dataclasses import dataclass
+
+from deprecate import deprecated
+
+
+@dataclass
+class Config:
+    timeout_ms: int = 30_000  # renamed from `timeout`
+
+    @property
+    @deprecated(deprecated_in="1.0", remove_in="2.0")
+    def timeout(self) -> int:
+        """Deprecated — use ``timeout_ms`` instead."""
+        return self.timeout_ms // 1000
+
+
+cfg = Config(timeout_ms=5_000)
+print(cfg.timeout)  # FutureWarning fired; prints 5
+```
+
+<details>
+  <summary>Output: <code>cfg.timeout</code></summary>
+
+```
+5
+```
+
+</details>
+
+**Read-write alias (warns on read and write):** use the outer order and chain `.setter`:
+
+```python
+from dataclasses import dataclass
+
+from deprecate import deprecated
+
+
+@dataclass
+class Config:
+    timeout_ms: int = 30_000  # renamed from `timeout`
+
+    @deprecated(deprecated_in="1.0", remove_in="2.0")
+    @property
+    def timeout(self) -> int:
+        return self.timeout_ms // 1000
+
+    @timeout.setter
+    def timeout(self, value: int) -> None:
+        self.timeout_ms = value * 1000
+
+
+cfg = Config(timeout_ms=5_000)
+print(cfg.timeout)  # FutureWarning fired; prints 5
+cfg.timeout = 10  # FutureWarning fired; sets timeout_ms = 10_000
+print(cfg.timeout_ms)  # prints 10_000
+```
+
+<details>
+  <summary>Output: <code>cfg.timeout; cfg.timeout_ms</code></summary>
+
+```
+5
+10000
+```
+
+</details>
+
+`cfg.timeout` fires `FutureWarning` (from `@deprecated`) and the getter body returns `cfg.timeout_ms // 1000`. `cfg.timeout = 5` fires `FutureWarning` and the setter body assigns `cfg.timeout_ms = 5000`.
+
+!!! warning "Do not shadow a dataclass field"
+
+    Do **not** use the same name as an existing dataclass field for the deprecated property. The `@dataclass`-generated `__init__` performs `self.field = value`, which conflicts with a property descriptor of the same name. Use a different name for the deprecated alias and keep the dataclass field under its new name.
+
+The same pattern works on regular (non-dataclass) classes — replace field access with `self._attr` lookups in the accessor body. `@deprecated` only adds the warning in either case.
 
-!!! tip "Prefer `@property @deprecated` (deprecated closer to `def`)"
+!!! note "`target=<callable>` not supported on properties"
 
-    The inner-first order is the conventional Python style. Follow this pattern for consistency if your team has no existing convention.
+    `@deprecated` rejects `target=<callable>` on a `property` with `TypeError`. Properties have three independent accessors (`fget`, `fset`, `fdel`); there is no single callable to forward to. Delegate in each accessor body as shown above.
 
 ## Deprecating generator functions