module Parallel
Overview
Core module for parallel processing functionality
Defined in:
parallel/context.crparallel/core.cr
parallel/version.cr
Constant Summary
-
VERSION =
{{ (`shards version /home/runner/work/parallel/parallel/src/parallel`).chomp.stringify }}
Class Method Summary
-
.adaptive_chunk_size(collection_size : Int32) : Int32
Determines optimal chunk size for adaptive chunking.
-
.check_empty_and_size(collection) : Tuple(Bool, Int32)
Unified empty check for collections Returns {is_empty, estimated_size}
-
.execution_context : Fiber::ExecutionContext::Parallel
Returns the current default context.
-
.execution_context=(context : Fiber::ExecutionContext::Parallel) : Fiber::ExecutionContext::Parallel
Sets the default context explicitly.
-
.log_fiber_exception(ex : Exception, task_info : String | Nil = nil)
Logs fiber exceptions with context information
-
.parallel_each(collection_size : Int32, context : Fiber::ExecutionContext::Parallel, chunk_size : Int32, &block : Int32 -> _)
Common parallel each implementation This method handles the core logic for both Enumerable and Indexable versions
-
.parallel_each_enumerable(enumerable : Enumerable(T), context : Fiber::ExecutionContext::Parallel, chunk_size : Int32, &block : T -> _) forall T
Lazy parallel each implementation for Enumerable collections This method processes elements without materializing the entire collection Uses lock-free work-stealing approach for better performance
-
.parallel_map_enumerable(enumerable : Enumerable(T), context : Fiber::ExecutionContext::Parallel, chunk_size : Int32, &block : T -> U) forall T, U
Common parallel map implementation for Enumerable collections using lazy evaluation This method handles the core logic for Enumerable versions without creating intermediate arrays Uses lock-free work-stealing approach for better performance
-
.parallel_map_indexable(collection_size : Int32, context : Fiber::ExecutionContext::Parallel, chunk_size : Int32, &block : Int32 -> U) forall U
Common parallel map implementation for Indexable collections This method handles the core logic for Indexable versions using direct index access
-
.validate_chunk_size(chunk : Int32 | Nil, collection_size : Int32) : Int32
Validates and normalizes chunk size parameter Returns the validated chunk size or raises an exception for invalid values
Class Method Detail
Determines optimal chunk size for adaptive chunking. The computed size is clamped to 1..1000.
Unified empty check for collections Returns {is_empty, estimated_size}
Returns the current default context. Lazily creates a dedicated context.
Sets the default context explicitly.
Logs fiber exceptions with context information
Common parallel each implementation This method handles the core logic for both Enumerable and Indexable versions
Lazy parallel each implementation for Enumerable collections This method processes elements without materializing the entire collection Uses lock-free work-stealing approach for better performance
Common parallel map implementation for Enumerable collections using lazy evaluation This method handles the core logic for Enumerable versions without creating intermediate arrays Uses lock-free work-stealing approach for better performance
Common parallel map implementation for Indexable collections This method handles the core logic for Indexable versions using direct index access
Validates and normalizes chunk size parameter Returns the validated chunk size or raises an exception for invalid values