Revision a121721f975fc4105ed24ebd0ad1020d08d07a38 authored by Shuhei Kadowaki on 01 November 2021, 10:49:07 UTC, committed by GitHub on 01 November 2021, 10:49:07 UTC
* inference: form `PartialStruct` for extra type information propagation

This commit forms `PartialStruct` whenever there is any type-level
refinement available about a field, even if it's not "constant" information.

In Julia "definitions" are allowed to be abstract whereas "usages"
(i.e. callsites) are often concrete. The basic idea is to allow inference
to make more use of such precise callsite type information by encoding it
as `PartialStruct`.

This may increase optimization possibilities of "unidiomatic" Julia code,
which may contain poorly-typed definitions, like this very contrived example:
```julia
struct Problem
    n; s; c; t
end

function main(args...)
    prob = Problem(args...)
    s = 0
    for i in 1:prob.n
        m = mod(i, 3)
        s += m == 0 ? sin(prob.s) : m == 1 ? cos(prob.c) : tan(prob.t)
    end
    return prob, s
end

main(10000, 1, 2, 3)
```

One of the obvious limitation is that this extra type information can be
propagated inter-procedurally only as a const-propagation.
I'm not sure this kind of "just a type-level" refinement can often make
constant-prop' successful (i.e. shape-up a method body and allow it to
be inlined, encoding the extra type information into the generated code),
thus I didn't not modify any part of const-prop' heuristics.

So the improvements from this change might not be very useful for general
inter-procedural analysis currently, but they should definitely improve the
accuracy of local analysis and very simple inter-procedural analysis.
1 parent 6c274ed
Raw File
generator.jl
# This file is a part of Julia. License is MIT: https://julialang.org/license

"""
    Generator(f, iter)

Given a function `f` and an iterator `iter`, construct an iterator that yields
the values of `f` applied to the elements of `iter`.
The syntax for constructing an instance of this type is `f(x) for x in iter [if cond(x)::Bool] `.
The `[if cond(x)::Bool]` expression is optional and acts as a "guard", effectively
filtering out values where the condition is false.

```jldoctest
julia> g = (abs2(x) for x in 1:5 if x != 3);

julia> for x in g
           println(x)
       end
1
4
16
25

julia> collect(g)
4-element Vector{Int64}:
  1
  4
 16
 25
```
"""
struct Generator{I,F}
    f::F
    iter::I
end

Generator(f, I1, I2, Is...) = Generator(a->f(a...), zip(I1, I2, Is...))

Generator(::Type{T}, iter::I) where {T,I} = Generator{I,Type{T}}(T, iter)

Generator(::Type{T}, I1, I2, Is...) where {T} = Generator(a->T(a...), zip(I1, I2, Is...))

function iterate(g::Generator, s...)
    @inline
    y = iterate(g.iter, s...)
    y === nothing && return nothing
    y = y::Tuple{Any, Any} # try to give inference some idea of what to expect about the behavior of the next line
    return (g.f(y[1]), y[2])
end

length(g::Generator) = length(g.iter)
size(g::Generator) = size(g.iter)
axes(g::Generator) = axes(g.iter)
ndims(g::Generator) = ndims(g.iter)
keys(g::Generator) = keys(g.iter)


## iterator traits

abstract type IteratorSize end
struct SizeUnknown <: IteratorSize end
struct HasLength <: IteratorSize end
struct HasShape{N} <: IteratorSize end
struct IsInfinite <: IteratorSize end

"""
    IteratorSize(itertype::Type) -> IteratorSize

Given the type of an iterator, return one of the following values:

* `SizeUnknown()` if the length (number of elements) cannot be determined in advance.
* `HasLength()` if there is a fixed, finite length.
* `HasShape{N}()` if there is a known length plus a notion of multidimensional shape (as for an array).
   In this case `N` should give the number of dimensions, and the [`axes`](@ref) function is valid
   for the iterator.
* `IsInfinite()` if the iterator yields values forever.

The default value (for iterators that do not define this function) is `HasLength()`.
This means that most iterators are assumed to implement [`length`](@ref).

This trait is generally used to select between algorithms that pre-allocate space for their
result, and algorithms that resize their result incrementally.

```jldoctest
julia> Base.IteratorSize(1:5)
Base.HasShape{1}()

julia> Base.IteratorSize((2,3))
Base.HasLength()
```
"""
IteratorSize(x) = IteratorSize(typeof(x))
IteratorSize(::Type) = HasLength()  # HasLength is the default

IteratorSize(::Type{<:Tuple}) = HasLength()
IteratorSize(::Type{<:AbstractArray{<:Any,N}})  where {N} = HasShape{N}()
IteratorSize(::Type{Generator{I,F}}) where {I,F} = IteratorSize(I)

IteratorSize(::Type{Any}) = SizeUnknown()

haslength(iter) = IteratorSize(iter) isa Union{HasShape, HasLength}

abstract type IteratorEltype end
struct EltypeUnknown <: IteratorEltype end
struct HasEltype <: IteratorEltype end

"""
    IteratorEltype(itertype::Type) -> IteratorEltype

Given the type of an iterator, return one of the following values:

* `EltypeUnknown()` if the type of elements yielded by the iterator is not known in advance.
* `HasEltype()` if the element type is known, and [`eltype`](@ref) would return a meaningful value.

`HasEltype()` is the default, since iterators are assumed to implement [`eltype`](@ref).

This trait is generally used to select between algorithms that pre-allocate a specific
type of result, and algorithms that pick a result type based on the types of yielded
values.

```jldoctest
julia> Base.IteratorEltype(1:5)
Base.HasEltype()
```
"""
IteratorEltype(x) = IteratorEltype(typeof(x))
IteratorEltype(::Type) = HasEltype()  # HasEltype is the default

IteratorEltype(::Type{Generator{I,T}}) where {I,T} = EltypeUnknown()

IteratorEltype(::Type{Any}) = EltypeUnknown()
back to top