futuresdr 0.6.0

An Experimental Async SDR Runtime for Heterogeneous Architectures.
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142

use std::future::Future;
use std::pin::Pin;

use crate::runtime::Result;
use crate::runtime::dev::BlockMeta;
use crate::runtime::dev::MessageOutputs;
use crate::runtime::dev::WorkIo;
use crate::runtime::kernel_interface::KernelInterface;

/// Send-capable marker for normal runtime blocks.
///
/// This keeps verbose return-type-notation bounds in one place: normal blocks
/// must have a `Send` value, `Send` block-on future, and `Send` futures returned
/// from the kernel lifecycle methods and generated block interface.
#[doc(hidden)]
pub trait SendKernel: Kernel<BlockOn: Send> + KernelInterface + Send
where
    Self: Kernel<work(..): Send, init(..): Send, deinit(..): Send>,
    Self: KernelInterface<stream_ports_notify_finished(..): Send, call_handler(..): Send>,
{
}

impl<T> SendKernel for T where
    T: Kernel<BlockOn: Send, work(..): Send, init(..): Send, deinit(..): Send>
        + KernelInterface<stream_ports_notify_finished(..): Send, call_handler(..): Send>
        + Send
{
}

/// Processing logic for a block.
///
/// `Kernel` is the central trait custom block authors implement. The
/// `#[derive(Block)]` macro declares stream and message ports from annotated
/// fields and methods; the `Kernel` implementation supplies initialization,
/// work, and shutdown behavior.
///
/// The runtime calls [`Kernel::init`] once, then repeatedly calls
/// [`Kernel::work`] until the block marks itself finished or the flowgraph is
/// stopped, and finally calls [`Kernel::deinit`]. A `work()` implementation
/// should consume and produce exactly the number of stream items it handled and
/// use [`WorkIo`] to request another immediate call or finish. Blocks that need
/// to wait on their own future can provide it through [`Kernel::block_on`].
///
/// Normal runtime entry points accept only kernels whose value, block-on future,
/// and returned futures are `Send`. Kernels that do not satisfy these bounds can
/// still run in a local domain.
///
/// ```
/// use futuresdr::runtime::dev::prelude::*;
///
/// #[derive(Block)]
/// struct Scale {
///     #[input]
///     input: DefaultCpuReader<f32>,
///     #[output]
///     output: DefaultCpuWriter<f32>,
///     gain: f32,
/// }
///
/// impl Kernel for Scale {
///     async fn work(
///         &mut self,
///         io: &mut WorkIo,
///         _mo: &mut MessageOutputs,
///         _meta: &BlockMeta,
///     ) -> Result<()> {
///         let input = self.input.slice();
///         let output = self.output.slice();
///         let input_len = input.len();
///         let n = input_len.min(output.len());
///
///         for i in 0..n {
///             output[i] = input[i] * self.gain;
///         }
///
///         self.input.consume(n);
///         self.output.produce(n);
///
///         if self.input.finished() && n == input_len {
///             io.finished = true;
///         }
///
///         Ok(())
///     }
/// }
/// ```
pub trait Kernel {
    /// Typed future that may be used to wake the block again.
    type BlockOn: Future<Output = ()> + 'static = std::future::Pending<()>;

    /// Return a typed future that may wake this block again.
    ///
    /// When `work()` does not request an immediate call and the block is not
    /// finished, the runtime waits for inbox/stream notifications. If this
    /// method returns `Some`, the runtime also awaits that future and calls
    /// `work()` again when either the future resolves or a notification arrives.
    /// Return `None` while no block-specific future should be polled.
    fn block_on(&mut self) -> Option<Pin<&mut Self::BlockOn>> {
        None
    }

    /// Process stream data and emit messages.
    ///
    /// Implementations inspect their input buffers, write output buffers, update
    /// consume/produce counts, optionally post PMTs through [`MessageOutputs`],
    /// and update [`WorkIo`] flags before returning.
    fn work(
        &mut self,
        _io: &mut WorkIo,
        _mo: &mut MessageOutputs,
        _b: &BlockMeta,
    ) -> impl Future<Output = Result<()>> {
        async { Ok(()) }
    }

    /// Initialize the kernel before normal work starts.
    ///
    /// This is the place to allocate runtime resources or send initial messages.
    /// Stream ports have already been initialized and validated when this method
    /// is called. Runtime metadata is read-only here; mutable block state should
    /// live in the kernel implementation.
    fn init(
        &mut self,
        _mo: &mut MessageOutputs,
        _b: &BlockMeta,
    ) -> impl Future<Output = Result<()>> {
        async { Ok(()) }
    }

    /// De-initialize the kernel after work has stopped.
    ///
    /// This is called during block shutdown even when the block stopped because
    /// the flowgraph was terminated. It should release resources owned by the
    /// block and may post final messages.
    fn deinit(
        &mut self,
        _mo: &mut MessageOutputs,
        _b: &BlockMeta,
    ) -> impl Future<Output = Result<()>> {
        async { Ok(()) }
    }
}