AutoBool allow_* flags by AdrianSosic · Pull Request #742 · emdgroup/baybe

AdrianSosic · 2026-02-10T08:52:38Z

Fixes #733.

The drawback of the previous approach was that the flag attributes had a union bool | UnspecifiedType type, which meant that they could not be safely accessed without additional conversion logic, resulting in the bug reported in #733.

To fix this, the user-facing value of the flags should always be a simple bool at query time, while still offering a clever "auto-determine" mechanism that does not require the user to explicitly set them depending on the search space type.

There are two options to achieve this:

Using a context-specific conversion at attribute setting time. While possible, it has the disadvantage that it complicates validation, which usually happens only after conversion.
For this reason, we use an AutoBool approach, where the user-specified value is stored as a private attribute and the public value is resolved lazily at query time.

Copilot

Pull request overview

Updates Campaign.allow_* flags to use an AutoBool mechanism so callers always observe plain bool values at query time, while still supporting an “auto” mode.

Changes:

Replaced bool | UnspecifiedType allow-flag storage with private AutoBool attrs fields plus public bool-returning properties.
Updated validation to raise IncompatibilityError for incompatible allow-flag settings when continuous subspaces are involved.
Refreshed tests and changelog to cover/describe the new behavior (including "auto").

Reviewed changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated 4 comments.

File	Description
tests/test_campaign.py	Updates parametrization and assertions to validate lazy resolution of allow flags and new error type.
baybe/campaign.py	Introduces `AutoBool`-backed private fields, new validator, and public `bool` properties for allow flags.
CHANGELOG.md	Documents the behavioral change and a cache-related fix.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

Copilot · 2026-02-10T08:55:43Z

tests/test_campaign.py

    if space_type is SearchSpaceType.DISCRETE:
        parameter = NumericalDiscreteParameter("p", [0, 1])
    else:
        parameter = NumericalContinuousParameter("p", [0, 1])


The test parametrizes space_type with SearchSpaceType.HYBRID, but the constructed parameter only yields either a discrete or continuous search space. As a result, the HYBRID case is not actually exercised (and Campaign.searchspace.type likely won’t be HYBRID). Consider constructing an actual hybrid search space for the HYBRID branch (e.g., by explicitly building a SearchSpace with both discrete and continuous subspaces) so the new HYBRID expectations are truly validated.

Copilot · 2026-02-10T08:55:43Z

baybe/campaign.py

+    _allow_recommending_already_measured: AutoBool = field(
+        alias="allow_recommending_already_measured",
+        default=AutoBool.AUTO,
+        converter=AutoBool.from_unstructured,
        validator=_validate_allow_flag,
        on_setattr=_set_with_cache_cleared,
        kw_only=True,
    )
-    """Allow to recommend experiments that were already measured earlier.
-    Can only be set for discrete search spaces."""
+    """Allow recommending experiments that have already been measured."""

-    allow_recommending_already_recommended: bool | UnspecifiedType = field(
-        default=Factory(
-            _make_allow_flag_default_factory(default=False), takes_self=True
-        ),
+    _allow_recommending_already_recommended: AutoBool = field(
+        alias="allow_recommending_already_recommended",
+        default=AutoBool.AUTO,
+        converter=AutoBool.from_unstructured,
        validator=_validate_allow_flag,
        on_setattr=_set_with_cache_cleared,
        kw_only=True,
    )
-    """Allow to recommend experiments that were already recommended earlier.
-    Can only be set for discrete search spaces."""
+    """Allow recommending experiments that have already been recommended."""

-    allow_recommending_pending_experiments: bool | UnspecifiedType = field(
-        default=Factory(
-            _make_allow_flag_default_factory(default=False), takes_self=True
-        ),
+    _allow_recommending_pending_experiments: AutoBool = field(
+        alias="allow_recommending_pending_experiments",
+        default=AutoBool.AUTO,
+        converter=AutoBool.from_unstructured,
        validator=_validate_allow_flag,
        on_setattr=_set_with_cache_cleared,
        kw_only=True,
    )


These attrs fields were renamed to private _allow_* names with public-facing alias=.... Any existing code that introspects attrs via fields(Campaign).allow_recommending_* will now break (those fields no longer exist as attrs attributes). Update such call sites to reference the new attrs fields (e.g., fields(Campaign)._allow_recommending_already_measured) and use .alias when you need the user-facing name for messages/config output.

Copilot · 2026-02-10T08:55:43Z

baybe/campaign.py

+        return bool(self._allow_recommending_already_measured)
+
+    @property
+    def allow_recommending_already_recommended(self) -> bool:
+        """Allow recommending experiments that have already been recommended."""
+        if self._allow_recommending_already_recommended is AutoBool.AUTO:
+            return False
+        return bool(self._allow_recommending_already_recommended)
+
+    @property
+    def allow_recommending_pending_experiments(self) -> bool:
+        """Allow recommending pending experiments."""
+        if self._allow_recommending_pending_experiments is AutoBool.AUTO:
+            return False
+        return bool(self._allow_recommending_pending_experiments)


These properties rely on bool(AutoBoolValue) for non-AUTO cases. That’s easy to misread and can be brittle depending on how AutoBool is implemented (e.g., Enum truthiness vs IntEnum vs custom __bool__). Consider making the mapping explicit (e.g., comparing to AutoBool.TRUE/AutoBool.FALSE or using a dedicated AutoBool.to_bool() helper) to keep semantics unambiguous.

Suggested change

return bool(self._allow_recommending_already_measured)

@property

def allow_recommending_already_recommended(self) -> bool:

"""Allow recommending experiments that have already been recommended."""

if self._allow_recommending_already_recommended is AutoBool.AUTO:

return False

return bool(self._allow_recommending_already_recommended)

@property

def allow_recommending_pending_experiments(self) -> bool:

"""Allow recommending pending experiments."""

if self._allow_recommending_pending_experiments is AutoBool.AUTO:

return False

return bool(self._allow_recommending_pending_experiments)

return self._allow_recommending_already_measured is AutoBool.TRUE

@property

def allow_recommending_already_recommended(self) -> bool:

"""Allow recommending experiments that have already been recommended."""

if self._allow_recommending_already_recommended is AutoBool.AUTO:

return False

return self._allow_recommending_already_recommended is AutoBool.TRUE

@property

def allow_recommending_pending_experiments(self) -> bool:

"""Allow recommending pending experiments."""

if self._allow_recommending_pending_experiments is AutoBool.AUTO:

return False

return self._allow_recommending_pending_experiments is AutoBool.TRUE

Copilot · 2026-02-10T08:55:44Z

CHANGELOG.md


+### Changed
+- The `Campaign.allow_*` flag mechanism is now based on `AutoBool` logic, providing
+  well-defined Boolean values at query time while exposing the `AUTO` option to the user


The changelog mentions an AUTO option, but the user-facing input in tests/messages appears to be the string 'auto' (and/or AutoBool.AUTO). Consider clarifying the exact supported user inputs (e.g., true/false/'auto' in JSON/configs, and/or AutoBool.AUTO in Python) so external users know what to pass.

Suggested change

well-defined Boolean values at query time while exposing the `AUTO` option to the user

well-defined Boolean values at query time while exposing an automatic mode to the user

(e.g., via the string `"auto"` in JSON/configs or `AutoBool.AUTO` in Python code)

AVHopp · 2026-02-10T10:35:02Z

@AdrianSosic: The create_from_config example seems to be failing, causing the docs fail

AVHopp

Except for the failing tests this looks good to me.

baybe/campaign.py

Scienfitz · 2026-02-16T14:14:54Z

baybe/campaign.py

is there anything in the doc that will make it easy to understand what the (finally resolved bollean) values are for the respective cases? Currently I think users would ahve to bother witht he logi. or execute code to underssand that,

Can we add a table? Or can the docstrings of the flags be expanded?

You mean something like a table that shows how the auto-values resolve for the different flags and search space types? I have nothing against that, but do you see any way how we can ensure that stuff stays in sync (especially given that meddling with the flags sort of has been our yearly fun exercise over and over again)?

well then lets have a static table, it wont change frequently anyway

Refactor Campaign.allow_* flags using AutoBool

77da6eb

AdrianSosic self-assigned this Feb 10, 2026

Copilot AI review requested due to automatic review settings February 10, 2026 08:52

AdrianSosic requested review from AVHopp and Scienfitz as code owners February 10, 2026 08:52

AdrianSosic added the refactor label Feb 10, 2026

Copilot AI reviewed Feb 10, 2026

View reviewed changes

AVHopp approved these changes Feb 10, 2026

View reviewed changes

baybe/campaign.py Show resolved Hide resolved

AdrianSosic added 2 commits February 10, 2026 16:44

Add hooks for AutoBool

8d4d3b1

Fix attribute references

be4b807

AdrianSosic mentioned this pull request Feb 10, 2026

Bugged recommendation cache check accessing allow_recommending_already_recommended when it's UNSPECIFIED #733

Open

Scienfitz reviewed Feb 16, 2026

View reviewed changes

Fix automatic value of allow_* flags

0e8ee75

Scienfitz approved these changes Mar 6, 2026

View reviewed changes

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

AutoBool allow_* flags#742

AutoBool allow_* flags#742
AdrianSosic wants to merge 4 commits intomainfrom
refactor/allow_flags

AdrianSosic commented Feb 10, 2026 •

edited

Loading

Uh oh!

Copilot AI left a comment

Uh oh!

Copilot AI Feb 10, 2026

Uh oh!

Copilot AI Feb 10, 2026

Uh oh!

Copilot AI Feb 10, 2026

Uh oh!

Copilot AI Feb 10, 2026

Uh oh!

AVHopp commented Feb 10, 2026

Uh oh!

AVHopp left a comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Scienfitz Feb 16, 2026

Uh oh!

AdrianSosic Feb 17, 2026

Uh oh!

Scienfitz Mar 5, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

	well-defined Boolean values at query time while exposing the `AUTO` option to the user
	well-defined Boolean values at query time while exposing an automatic mode to the user
	(e.g., via the string `"auto"` in JSON/configs or `AutoBool.AUTO` in Python code)

Conversation

AdrianSosic commented Feb 10, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

Copilot AI Feb 10, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Feb 10, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Feb 10, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Feb 10, 2026

Choose a reason for hiding this comment

Uh oh!

AVHopp commented Feb 10, 2026

Uh oh!

AVHopp left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Scienfitz Feb 16, 2026

Choose a reason for hiding this comment

Uh oh!

AdrianSosic Feb 17, 2026

Choose a reason for hiding this comment

Uh oh!

Scienfitz Mar 5, 2026

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

AdrianSosic commented Feb 10, 2026 •

edited

Loading