Mavryk_stdlib.Circular_buffer
This module implements a bufferisation abstraction to store temporary raw data chunks (as bytes) when chunks are read sequentially. The function write
allows to store chunks in the buffer and the function read to read them from the buffer.
The global contract is that if we write consecutively d1;d2
onto the buffer, then we have to fully read d1
and d2
, in that order.
This contract is not enforced by the library, it is the user responsibility to respect it.
If the circular buffer is full, a new temporary buffer is allocated to store the chunk of data to be written.
val create : ?maxlength:int -> ?fresh_buf_size:int -> unit -> t
create ?maxlength ?fresh_buf_size ()
creates a buffer of size maxlength
(by default 32
kb). If the buffer is full, a buffer of size fresh_buf_size
is allocated (by default 2
kb).
val write :
maxlen:int ->
fill_using:
(Stdlib.Bytes.t -> int -> int -> (int, 'error) Stdlib.result Lwt.t) ->
t ->
(data, 'error) Stdlib.result Lwt.t
write ~maxlen ~fill_using buffer
calls fill_using buf offset
maxlen
where buf
is a buffer that has room for maxlen
data starting from offset
.
Assumes that fill_using
returns the exact amount of written bytes.
Behaviour is unspecified if fill_using
writes more than maxlen
data or lies on the number of written bytes.
It returns a data descriptor for the supposedly written chunk.
read data ~len ~into:buf buffer ~offset
copies len
data from the data
chunk into buf
. If len
is not provided, it copies all the data. If len
is less than the amount of data available, it returns a new handler of the remainder.
data
has been produced by a write
attempt in buffer
.len
is less than length data
.val length : data -> int
length data
returns the amount of available bytes in data