Backports release 1.11 #53790
Backports release 1.11 #53790
Conversation
Also, simplifies the docstring by splitting the usage into two lines. (cherry picked from commit 0b1587c)
(cherry picked from commit 1d532f4)
~~Apparently somewhere in codegen inside `emit_codeinst`some piece of code is relying on the implicit try catch gcstack restoring. I haven't got the analyzer working on that file yet (it has hundreds of issues and it doesn't like C++ code that much + the file is tens of thousands of lines after includes so it struggles).~~ This fixes the compileall segfault in apple-aarch64 ci. #53811 (cherry picked from commit 52fc796)
Currently it's a bit excessive in the `Pkg.test` precompile job  This PR ``` Precompiling project for configuration --code-coverage=none --color=yes --check-bounds=yes --warn-overwrite=yes --depwarn=yes --inline=yes --startup-file=no --track-allocation=none... 354.9 ms ✓ RFFT 1 dependency successfully precompiled in 1 seconds. 38 already precompiled. ``` Pkg could also just set the non-default flags to minimize the list. (cherry picked from commit 9291845)
Fixes JuliaDebug/Cthulhu.jl#541 by picking up libuv/libuv@3ecce91. (cherry picked from commit 5ed51d3)
MPFR 4.2.1 was released on 22 August 2023. It contains bugfixes. (cherry picked from commit 61caaa8)
(cherry picked from commit fc2d3af)
…3383) With this, the following (and equivalent calls) work: ```julia julia> copyto!(view(zeros(BigInt, 2), 1:2), Vector{BigInt}(undef,2)) 2-element view(::Vector{BigInt}, 1:2) with eltype BigInt: #undef #undef julia> copyto!(view(zeros(BigInt, 2), 1:2), view(Vector{BigInt}(undef,2), 1:2)) 2-element view(::Vector{BigInt}, 1:2) with eltype BigInt: #undef #undef ``` Close #53098. With this, all the `_unsetindex!` branches in `copyto_unaliased!` work for `Array`-views, and this makes certain indexing operations vectorize and speed-up: ```julia julia> using BenchmarkTools julia> a = view(rand(100,100), 1:100, 1:100); b = view(similar(a), axes(a)...); julia> @Btime copyto!($b, $a); 16.427 μs (0 allocations: 0 bytes) # master 2.308 μs (0 allocations: 0 bytes) # PR ``` Improves (but doesn't resolve) #40962 and #53158 ```julia julia> a = rand(40,40); b = rand(40,40); julia> @Btime $a[1:end,1:end] .= $b; 5.383 μs (0 allocations: 0 bytes) # v"1.12.0-DEV.16" 3.194 μs (0 allocations: 0 bytes) # PR ``` ƒ Co-authored-by: Jameson Nash <[email protected]> (cherry picked from commit 1a90409)
These had been copy-pasted incorrectly, and should throw an error if the indices are out of bounds. (cherry picked from commit 98b3f72)
|
@nanosoldier |
|
The package evaluation job you requested has completed - possible new issues were detected. |
Not sure why this typo didn't mess up the CI, but it looks like a clear problem, let's correct it. I'd appreciate any idea on how to exercise this change. - fixes JuliaDebug/JuliaInterpreter.jl#621
This PR adds a utility function `isrelocatable(pkg)` that can be used to check if `pkg` is already precompiled and if the associated cachefile is relocatable. The reason to implicitly perform the `isprecompiled` check is that the exact same computation needs to be done to find the right `.ji`. A `pkg` is said to be relocatable if 1. all `include()` paths are relocatable (they start with `@depot`), 2. all `include_dependency()` paths are relocatable (they start with `@depot` and `track_content=true` was used to include them). (cherry picked from commit e9d25ca)
(cherry picked from commit 57bbff6)
This reduces dynamic dispatch and makes JET happier.
Testing on v1.11:
```julia
julia> import LinearAlgebra: checksquare
julia> using JET
julia> @report_opt checksquare(rand(2,2), rand(2,2))
═════ 4 possible errors found ═════
┌ checksquare(::Matrix{Float64}, ::Matrix{Float64}) @ LinearAlgebra /cache/build/builder-amdci4-1/julialang/julia-release-1-dot-11/usr/share/julia/stdlib/v1.11/LinearAlgebra/src/LinearAlgebra.jl:307
│┌ string(::String, ::Tuple{Int64, Int64}) @ Base ./strings/io.jl:189
││┌ print_to_string(::String, ::Tuple{Int64, Int64}) @ Base ./strings/io.jl:150
│││┌ _unsafe_take!(io::IOBuffer) @ Base ./iobuffer.jl:494
││││┌ wrap(::Type{Array}, m::MemoryRef{UInt8}, l::Int64) @ Base ./array.jl:3101
│││││ failed to optimize due to recursion: wrap(::Type{Array}, ::MemoryRef{UInt8}, ::Int64)
││││└────────────────────
│││┌ print_to_string(::String, ::Vararg{Any}) @ Base ./strings/io.jl:143
││││ runtime dispatch detected: Base._str_sizehint(%17::Any)::Int64
│││└────────────────────
│││┌ print_to_string(::String, ::Vararg{Any}) @ Base ./strings/io.jl:148
││││ runtime dispatch detected: print(%59::IOBuffer, %97::Any)::Any
│││└────────────────────
│││┌ string(::String, ::Int64, ::String, ::Tuple{Int64}, ::String, ::Int64, ::String, ::Int64, ::String) @ Base ./strings/io.jl:189
││││ failed to optimize due to recursion: string(::String, ::Int64, ::String, ::Tuple{Int64}, ::String, ::Int64, ::String, ::Int64, ::String)
│││└────────────────────
julia> function checksquare(A...) # This PR
sizes = Int[]
for a in A
size(a,1)==size(a,2) || throw(DimensionMismatch(lazy"matrix is not square: dimensions are $(size(a))"))
push!(sizes, size(a,1))
end
return sizes
end
checksquare (generic function with 2 methods)
julia> @report_opt checksquare(rand(2,2), rand(2,2))
No errors detected
```
(cherry picked from commit d505c8c)
On `1.11.0-alpha2` Old: ```julia @benchmark Base.dec($0x1, $0, $false) BenchmarkTools.Trial: 10000 samples with 994 evaluations. Range (min … max): 33.702 ns … 4.242 μs ┊ GC (min … max): 0.00% … 97.61% Time (median): 37.626 ns ┊ GC (median): 0.00% Time (mean ± σ): 45.787 ns ± 147.794 ns ┊ GC (mean ± σ): 14.53% ± 4.47% ▄▅▆▇█▇▇▅▃▃▂▂▂▁ ▁▂▁▁▁ ▁▁ ▁ ▂ ▄▇███████████████▇▇██████▇█▆▆▄▄▃▄▅▄▆▇████████▆▅▅▇▆▅▆▄▄▅▄▄▄▁▅ █ 33.7 ns Histogram: log(frequency) by time 67.5 ns < Memory estimate: 88 bytes, allocs estimate: 3. ``` New: ```julia BenchmarkTools.Trial: 10000 samples with 995 evaluations. Range (min … max): 27.538 ns … 3.397 μs ┊ GC (min … max): 0.00% … 97.86% Time (median): 30.151 ns ┊ GC (median): 0.00% Time (mean ± σ): 34.547 ns ± 105.101 ns ┊ GC (mean ± σ): 10.37% ± 3.39% ▁ █▆▃ ▁ ▂▂▃▃▅█████▆████▆▄▄▃▃▃▃▃▃▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▂▁▂▂▂▂▂▁▂▂▂▂▂▂▂▂▂▂▂▂▂ ▃ 27.5 ns Histogram: frequency by time 43.8 ns < Memory estimate: 56 bytes, allocs estimate: 2. ``` Fixes #53950, actually now even faster than `1.10.2`. It doesn't look like the length is ever changed and we don't return these `StringMemory`s so this change should be fine. (cherry picked from commit 0e59c9e)
(cherry picked from commit d963a34)
…files (#53905) Fixes #53859 (comment), which was actually fixed before in #52346, but #52750 undid that fix. This PR does not directly address #53859, because I can not reproduce it atm. --- The `@depot` resolution logic for `include()` files is adjusted as follows: 1. (new behavior) If the cache is not relocatable because of an absolute path, we ignore that path for the depot search. Recompilation will be triggered by `stale_cachefile()` if that absolute path does not exist. Previously this caused any `@depot` tags to be not resolved and so trigger recompilation. 2. (new behavior) If we can't find a depot for a relocatable path, we still replace it with the depot we found from other files. Recompilation will be triggered by `stale_cachefile()` because the resolved path does not exist. 3. (this behavior is kept) We require that relocatable paths all resolve to the same depot. 4. (new behavior) We no longer use the first matching depot for replacement, but instead we explicitly check that all resolve to the same depot. This has two reasons: - We want to scan all source files anyways in order to provide logs for 1. and 2. above, so the check is free. - It is possible that a depot might be missing source files. Assume that we have two depots on `DEPOT_PATH`, `depot_complete` and `depot_incomplete`. If `DEPOT_PATH=["depot_complete","depot_incomplete"]` then no recompilation shall happen, because `depot_complete` will be picked. If `DEPOT_PATH=["depot_incomplete","depot_complete"]` we trigger recompilation and hopefully a meaningful error about missing files is thrown. If we were to just select the first depot we find, then whether recompilation happens would depend on whether the first relocatable file resolves to `depot_complete` or `depot_incomplete`. (cherry picked from commit d8d3842)
- Make reshape and view with one based indexing on Memory produce Arrays - delete wrap Implements #53552 (comment) --------- Co-authored-by: Jameson Nash <[email protected]> (cherry picked from commit 273d91e)
Backported PRs:
_unsetindex!methods forSubArrays andCartesianIndexes #53383Base.isrelocatable(pkg)#53906LazyStringinLinearAlgebra.checksquareerror message #53961Allocs.@profiledoc string #53825LazyStringinDimensionMismatcherror messages in broadcasting #53975view(::Memory, ::Union{UnitRange, Base.OneTo})#53991Non-merged PRs with backport label:
jl_getaffinityandjl_setaffinity#53402