juce::AbstractFifo Class Reference

#include <juce_AbstractFifo.h>

Classes
class	ScopedReadWrite

Public Types
using	ScopedRead = ScopedReadWrite<ReadOrWrite::read>
using	ScopedWrite = ScopedReadWrite<ReadOrWrite::write>

Public Member Functions
	AbstractFifo (int capacity) noexcept
int	getTotalSize () const noexcept
int	getFreeSpace () const noexcept
int	getNumReady () const noexcept
void	reset () noexcept
void	setTotalSize (int newSize) noexcept
void	prepareToWrite (int numToWrite, int &startIndex1, int &blockSize1, int &startIndex2, int &blockSize2) const noexcept
void	finishedWrite (int numWritten) noexcept
void	prepareToRead (int numWanted, int &startIndex1, int &blockSize1, int &startIndex2, int &blockSize2) const noexcept
void	finishedRead (int numRead) noexcept
ScopedRead	read (int numToRead) noexcept
ScopedWrite	write (int numToWrite) noexcept

Private Types
enum class	ReadOrWrite { read , write }

Private Attributes
int	bufferSize
Atomic< int >	validStart
Atomic< int >	validEnd

Detailed Description

Encapsulates the logic required to implement a lock-free FIFO.

This class handles the logic needed when building a single-reader, single-writer FIFO.

It doesn't actually hold any data itself, but your FIFO class can use one of these to manage its position and status when reading or writing to it.

To use it, you can call prepareToWrite() to determine the position within your own buffer that an incoming block of data should be stored, and prepareToRead() to find out when the next outgoing block should be read from.

e.g.

struct MyFifo
{
    void addToFifo (const int* someData, int numItems)
    {
        const auto scope = abstractFifo.write (numItems);
 
        if (scope.blockSize1 > 0)
            copySomeData (myBuffer + scope.startIndex1, someData, scope.blockSize1);
 
        if (scope.blockSize2 > 0)
            copySomeData (myBuffer + scope.startIndex2, someData, scope.blockSize2);
    }
 
    void readFromFifo (int* someData, int numItems)
    {
        const auto scope = abstractFifo.read (numItems);
 
        if (scope.blockSize1 > 0)
            copySomeData (someData, myBuffer + scope.startIndex1, scope.blockSize1);
 
        if (scope.blockSize2 > 0)
            copySomeData (someData + scope.blockSize1, myBuffer + scope.startIndex2, scope.blockSize2);
    }
 
    AbstractFifo abstractFifo { 1024 };
    int myBuffer[1024];
};

@tags{Core}

Member Typedef Documentation

ScopedRead

using juce::AbstractFifo::ScopedRead = ScopedReadWrite<ReadOrWrite::read>

ScopedWrite

using juce::AbstractFifo::ScopedWrite = ScopedReadWrite<ReadOrWrite::write>

Member Enumeration Documentation

ReadOrWrite

enum class juce::AbstractFifo::ReadOrWrite

strongprivate

Enumerator
read
write

Constructor & Destructor Documentation

AbstractFifo()

juce::AbstractFifo::AbstractFifo ( int capacity )

noexcept

Creates a FIFO to manage a buffer with the specified capacity.

Member Function Documentation

finishedRead()

void juce::AbstractFifo::finishedRead ( int numRead )

noexcept

Called after reading from the FIFO, to indicate that this many items have now been consumed.

See also

prepareToRead

finishedWrite()

void juce::AbstractFifo::finishedWrite ( int numWritten )

noexcept

Called after writing from the FIFO, to indicate that this many items have been added.

See also

prepareToWrite

getFreeSpace()

int juce::AbstractFifo::getFreeSpace ( ) const

noexcept

Returns the number of items that can currently be added to the buffer without it overflowing.

getNumReady()

int juce::AbstractFifo::getNumReady ( ) const

noexcept

Returns the number of items that can currently be read from the buffer.

getTotalSize()

int juce::AbstractFifo::getTotalSize ( ) const

noexcept

Returns the total size of the buffer being managed.

prepareToRead()

void juce::AbstractFifo::prepareToRead ( int numWanted, int & startIndex1, int & blockSize1, int & startIndex2, int & blockSize2 ) const

noexcept

Returns the location within the buffer from which the next block of data should be read.

Because the section of data that you want to read from the buffer may overlap the end and wrap around to the start, two blocks within your buffer are returned, and you should read from both of them.

If the number of items you ask for is greater than the amount of data available, then blockSize1 + blockSize2 may add up to a lower value than numWanted. If this happens, you may decide to keep waiting and re-trying the method until there's enough data available.

After calling this method, if you choose to read the data, you must call finishedRead() to tell the FIFO how much data you have consumed.

e.g.

void readFromFifo (int* someData, int numItems)
{
    int start1, size1, start2, size2;
    prepareToRead (numSamples, start1, size1, start2, size2);
 
    if (size1 > 0)
        copySomeData (someData, myBuffer + start1, size1);
 
    if (size2 > 0)
        copySomeData (someData + size1, myBuffer + start2, size2);
 
    finishedRead (size1 + size2);
}

Parameters

numWanted	indicates how many items you'd like to add to the buffer
startIndex1	on exit, this will contain the start index in your buffer at which your data should be written
blockSize1	on exit, this indicates how many items can be written to the block starting at startIndex1
startIndex2	on exit, this will contain the start index in your buffer at which any data that didn't fit into the first block should be written
blockSize2	on exit, this indicates how many items can be written to the block starting at startIndex2

See also

finishedRead

prepareToWrite()

void juce::AbstractFifo::prepareToWrite ( int numToWrite, int & startIndex1, int & blockSize1, int & startIndex2, int & blockSize2 ) const

noexcept

Returns the location within the buffer at which an incoming block of data should be written.

Because the section of data that you want to add to the buffer may overlap the end and wrap around to the start, two blocks within your buffer are returned, and you should copy your data into the first one, with any remaining data spilling over into the second.

If the number of items you ask for is too large to fit within the buffer's free space, then blockSize1 + blockSize2 may add up to a lower value than numToWrite. If this happens, you may decide to keep waiting and re-trying the method until there's enough space available.

After calling this method, if you choose to write your data into the blocks returned, you must call finishedWrite() to tell the FIFO how much data you actually added.

e.g.

void addToFifo (const int* someData, int numItems)
{
    int start1, size1, start2, size2;
    prepareToWrite (numItems, start1, size1, start2, size2);
 
    if (size1 > 0)
        copySomeData (myBuffer + start1, someData, size1);
 
    if (size2 > 0)
        copySomeData (myBuffer + start2, someData + size1, size2);
 
    finishedWrite (size1 + size2);
}

Parameters

numToWrite	indicates how many items you'd like to add to the buffer
startIndex1	on exit, this will contain the start index in your buffer at which your data should be written
blockSize1	on exit, this indicates how many items can be written to the block starting at startIndex1
startIndex2	on exit, this will contain the start index in your buffer at which any data that didn't fit into the first block should be written
blockSize2	on exit, this indicates how many items can be written to the block starting at startIndex2

See also

finishedWrite

read()

AbstractFifo::ScopedRead juce::AbstractFifo::read ( int numToRead )

noexcept

Replaces prepareToRead/finishedRead with a single function. This function returns an object which contains the start indices and block sizes, and also automatically finishes the read operation when it goes out of scope.

{
    auto readHandle = fifo.read (4);
 
    for (auto i = 0; i != readHandle.blockSize1; ++i)
    {
        // read the item at index readHandle.startIndex1 + i
    }
 
    for (auto i = 0; i != readHandle.blockSize2; ++i)
    {
        // read the item at index readHandle.startIndex2 + i
    }
} // readHandle goes out of scope here, finishing the read operation

reset()

void juce::AbstractFifo::reset ( )

noexcept

Clears the buffer positions, so that it appears empty.

setTotalSize()

void juce::AbstractFifo::setTotalSize ( int newSize )

noexcept

Changes the buffer's total size. Note that this isn't thread-safe, so don't call it if there's any danger that it might overlap with a call to any other method in this class!

write()

AbstractFifo::ScopedWrite juce::AbstractFifo::write ( int numToWrite )

noexcept

Replaces prepareToWrite/finishedWrite with a single function. This function returns an object which contains the start indices and block sizes, and also automatically finishes the write operation when it goes out of scope.

{
    auto writeHandle = fifo.write (5);
 
    for (auto i = 0; i != writeHandle.blockSize1; ++i)
    {
        // write the item at index writeHandle.startIndex1 + i
    }
 
    for (auto i = 0; i != writeHandle.blockSize2; ++i)
    {
        // write the item at index writeHandle.startIndex2 + i
    }
} // writeHandle goes out of scope here, finishing the write operation

Member Data Documentation

bufferSize

int juce::AbstractFifo::bufferSize

private

validEnd

Atomic<int> juce::AbstractFifo::validEnd

private

validStart

Atomic<int> juce::AbstractFifo::validStart

private

---

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

• /home/runner/work/lmms-fork/lmms-fork/plugins/CarlaBase/carla/source/modules/juce_core/containers/juce_AbstractFifo.h
• /home/runner/work/lmms-fork/lmms-fork/plugins/CarlaBase/carla/source/modules/juce_core/containers/juce_AbstractFifo.cpp