This template class provides a concurrent registry for tasks that handles hash conflicts by allowing multiple task entries to coexist in the same hash bucket. More...

#include <task_registry_multi.h>

Classes
struct	Bucket
	Represents a bucket, capable of handling many subentries. More...

struct	Sub_entry
	Each bucket is capable of handling many subentries. More...

Public Types
using	Task_object = Task_object_type

using	Mutex = concurrency::Spin_lock_mutex

Public Member Functions
	Task_registry_multi (std::size_t capacity=default_capacity)
	Constructs buckets using defined capacity. More...

	~Task_registry_multi ()=default
	Destruct registry, deinitialize if applicable. More...

template<typename T >
bool	apply (Task_id_type id, T &&func)
	Calls 'func' on a given task object if active. More...

bool	activate (Task_id_type id, Task_object &&obj)
	Registers and activates a task. More...

bool	deactivate (Task_id_type id)
	Deactivates a task. More...

bool	bucket_active (Task_id_type id) const
	Checks if the bucket for the given task ID is active (contains any tasks) More...

Static Public Attributes
static constexpr std::size_t	default_capacity = 16384

Private Types
using	Bucket_ptr = std::unique_ptr< Bucket >

Private Member Functions
Bucket_ptr &	get_bucket (Task_id_type id)
	Obtains the bucket handle handling a given id, for internal use. More...

Private Attributes
std::vector< Bucket_ptr >	m_buckets
	Buckets, synchronized independently. More...

Detailed Description

template<typename Task_id_type, typename Task_object_type>
class mysql::scheduler::Task_registry_multi< Task_id_type, Task_object_type >

This template class provides a concurrent registry for tasks that handles hash conflicts by allowing multiple task entries to coexist in the same hash bucket.

It implements thread-safe access to task objects using a bucketed hash table with locks per bucket.

Threads do not contend when they operate on separate buckets. When a hash conflict occurs, threads use spin locks to lock requested resources (bucket).

Spin locks excel when locks are held for very brief periods, such as in this registry, where operations like checking task activation or updating entries are quick atomic operations. The spin lock avoids expensive context switches that std::mutex would incur.

Each bucket (determined by task ID hash) contains an unordered_map mapping task IDs to their associated objects and state. This design resolves collisions that could occur with simple hashing, ensuring that tasks with IDs mapping to the same bucket index can be stored and accessed efficiently.

Key features:

Hash conflict resolution: Multiple tasks per bucket using std::unordered_map.
Thread-safe: Each bucket has its own mutex.
Limited API: Only essential operations for task management.
Template-based: Supports various task ID and object types.

Usage example:

Task_registry_multi<Task_id, Task_info> registry(1000);
registry.activate(task_id, Task_info{...});
registry.apply(task_id, [](Task_info& info) { /* operate on info */ });
registry.deactivate(task_id);

Member Typedef Documentation

◆ Bucket_ptr

template<typename Task_id_type , typename Task_object_type >

using mysql::scheduler::Task_registry_multi< Task_id_type, Task_object_type >::Bucket_ptr = std::unique_ptr<Bucket>

private

◆ Mutex

template<typename Task_id_type , typename Task_object_type >

using mysql::scheduler::Task_registry_multi< Task_id_type, Task_object_type >::Mutex = concurrency::Spin_lock_mutex

◆ Task_object

template<typename Task_id_type , typename Task_object_type >

using mysql::scheduler::Task_registry_multi< Task_id_type, Task_object_type >::Task_object = Task_object_type

Constructor & Destructor Documentation

◆ Task_registry_multi()

template<typename Task_id_type , typename Task_object_type >

mysql::scheduler::Task_registry_multi< Task_id_type, Task_object_type >::Task_registry_multi ( std::size_t capacity = default_capacity )

inline

Constructs buckets using defined capacity.

◆ ~Task_registry_multi()

template<typename Task_id_type , typename Task_object_type >

mysql::scheduler::Task_registry_multi< Task_id_type, Task_object_type >::~Task_registry_multi ( )

default

Destruct registry, deinitialize if applicable.

Member Function Documentation

◆ activate()

template<typename Task_id_type , typename Task_object_type >

bool mysql::scheduler::Task_registry_multi< Task_id_type, Task_object_type >::activate	(	Task_id_type	id,
		Task_object &&	obj
	)

inline

Registers and activates a task.

Parameters

id	Task id
obj	Task object

Returns: true if the previous state was inactive (i.e., the task was newly registered), false if the previous state was active (task already present)

Exceptions

std::bad_alloc If memory allocation for the new entry fails

◆ apply()

template<typename Task_id_type , typename Task_object_type >

template<typename T >

bool mysql::scheduler::Task_registry_multi< Task_id_type, Task_object_type >::apply	(	Task_id_type	id,
		T &&	func
	)

inline

Calls 'func' on a given task object if active.

Parameters

id	Task id
func	function to be applied on a given registered task

Returns: true if task was found and active, false otherwise

◆ bucket_active()

template<typename Task_id_type , typename Task_object_type >

bool mysql::scheduler::Task_registry_multi< Task_id_type, Task_object_type >::bucket_active ( Task_id_type id ) const

inline

Checks if the bucket for the given task ID is active (contains any tasks)

Parameters

id Task id

Returns: true if bucket has at least one active task, false otherwise

◆ deactivate()

template<typename Task_id_type , typename Task_object_type >

bool mysql::scheduler::Task_registry_multi< Task_id_type, Task_object_type >::deactivate ( Task_id_type id )

inline

Deactivates a task.

Parameters

id Task id

Returns: true if deactivated, false if not found or not active

◆ get_bucket()

template<typename Task_id_type , typename Task_object_type >

Bucket_ptr & mysql::scheduler::Task_registry_multi< Task_id_type, Task_object_type >::get_bucket ( Task_id_type id )

inlineprivate

Obtains the bucket handle handling a given id, for internal use.

Parameters

id Task id

Returns: Bucket handle

Member Data Documentation

◆ default_capacity

template<typename Task_id_type , typename Task_object_type >

constexpr std::size_t mysql::scheduler::Task_registry_multi< Task_id_type, Task_object_type >::default_capacity = 16384

staticconstexpr

◆ m_buckets

template<typename Task_id_type , typename Task_object_type >

std::vector<Bucket_ptr> mysql::scheduler::Task_registry_multi< Task_id_type, Task_object_type >::m_buckets

private

Buckets, synchronized independently.

The documentation for this class was generated from the following file:

libs/mysql/scheduler/task_registry_multi.h

Classes

Public Types

Public Member Functions

Static Public Attributes

Private Types

Private Member Functions

Private Attributes

Detailed Description

Member Typedef Documentation

◆ Bucket_ptr

◆ Mutex

◆ Task_object

Constructor & Destructor Documentation

◆ Task_registry_multi()

◆ ~Task_registry_multi()

Member Function Documentation

◆ activate()

◆ apply()

◆ bucket_active()

◆ deactivate()

◆ get_bucket()

Member Data Documentation

◆ default_capacity

◆ m_buckets