Skip to main content

Browse the archive

swh:1:snp:a72e953ecd624a7df6e6196bbdd05851996c5e40
sort by:
Revision Author Date Message Commit Date
48dee43 Keno Fischer 05 August 2022, 08:16:20 UTC WIP 05 August 2022, 08:16:20 UTC
1996fb9 Keno Fischer 05 August 2022, 06:52:16 UTC WIP: ASAN revival 05 August 2022, 06:52:24 UTC
b503480 Keno Fischer 04 August 2022, 23:26:03 UTC Work around dlopen not working properly under sanitizer instrumentation `dlopen` has a mis-feature where it looks at the return address to determine the calling object to look at it's RUNPATH. Because asan intercepts `dlopen`, the calling object check finds asan rather than julia, causing an incorrect RUNPATH (and other flags to be used). Arguably, this is mostly a libc problem, because there isn't really a way to directly specify the resolution scope. I have sent a proposal to libc-coord [1] to fix this, but of course, we can't wait for that to percolate down to us. Instead, this takes advantage of the fact that almost all of our dlopen calls go through a single entrypoint in jl_dlopen, so we can insert additional logic here to make this work. This doesn't catch uses of `dlopen` in jlls (which is a problem for things like plugin loading in various jlls), but it at least makes base julia work. We can punt the jll question to another day - either with a patched libc in PkgEval or by patching the jll source with an analogous patch. Regardless, with this, Julia bootstraps properly under asan, without any special LD_LIBRARY_PATH hacks. [1] https://www.openwall.com/lists/libc-coord/2022/08/04/1 05 August 2022, 06:52:24 UTC
cf9489b Keno Fischer 03 August 2022, 23:46:08 UTC flisp: Fix memory leaks There's two independent issues here: 1. The table allocator assumes that small tables will be stored inline and do not need a finalizer. This is mostly true, except that hash collisions can cause premature growing of the inline table, so even for relatively small tables, we need to validate that the storage was not allocated out-of-line. 2. It is unsafe to clear the vtable finalizer pointer during the table allocation to supress the `add_finalizer` call. This is because the allocation of the table object itself may trigger GC of a different table, and without the finalizer set in the vtable, freeing of that table's memory space would get skipped. 05 August 2022, 06:52:24 UTC
01c0778 Shuhei Kadowaki 05 August 2022, 06:31:38 UTC effects: add effects analysis for array construction (#46015) This commit implements an infrastructure to analyze effects of `:foreigncall` expression, and especially implements effects analysis for array constructions. Some improvements: ```julia julia> @noinline construct_array(@nospecialize(T), args...) = Array{T}(undef, args...); julia> function check_dims(T, dims) construct_array(T, dims...) return nothing end; ``` ```julia julia> code_typed() do check_dims(Int, (1,2,3)) end ``` ```diff diff --git a/_master b/_pr index b0ed0eaac1..38e2d3553d 100644 --- a/_master +++ b/_pr @@ -1,6 +1,4 @@ 1-element Vector{Any}: CodeInfo( -1 ─ invoke Main.construct_array(Int64::Any, -1::Int64, 2::Vararg{Int64}, 3)::Any -│ %2 = Main.nothing::Nothing -└── return %2 +1 ─ return nothing ) => Nothing ``` ```julia julia> code_typed() do check_dims(Int, (-1,2,3)) end ``` ```diff diff --git a/_master b/_pr index b0ed0eaac1..e5e724db10 100644 --- a/_master +++ b/_pr @@ -1,6 +1,5 @@ 1-element Vector{Any}: CodeInfo( -1 ─ invoke Main.construct_array(Int64::Any, -1::Int64, 2::Vararg{Int64}, 3)::Any -│ %2 = Main.nothing::Nothing -└── return %2 -) => Nothing +1 ─ invoke Main.check_dims(Main.Int::Type, (-1, 2, 3)::Tuple{Int64, Int64, Int64})::Union{} +└── unreachable +) => Union{} ``` 05 August 2022, 06:31:38 UTC
e2bddbf Shuhei Kadowaki 05 August 2022, 03:46:04 UTC mark fastmath operations as not `:consistent` (#46143) We also need to mark `muladd` as not IPO-`:consistent, but it requires to revive #31193 to preserve the currently available optimizations so I left it as TODO for now. 05 August 2022, 03:46:04 UTC
de375ef Patrick White 04 August 2022, 19:21:45 UTC Fix dead link in 'Variables' section of the manual (#46248) 04 August 2022, 19:21:45 UTC
2c224e1 t-bltg 04 August 2022, 19:17:05 UTC fix `hvncat` docstring (#46017) * fix `hvncat` docstring * use `\Rightarrow` instead of `\implies` 04 August 2022, 19:17:05 UTC
bc4bb1b Oscar Smith 04 August 2022, 05:06:25 UTC make exp nothrow (#46238) This change allows our compiler to remove dead calls of these math ops. Co-authored-by: Simeon Schaub <simeondavidschaub99@gmail.com> Co-authored-by: Shuhei Kadowaki <40514306+aviatesk@users.noreply.github.com> 04 August 2022, 05:06:25 UTC
480df09 Shuhei Kadowaki 04 August 2022, 05:04:27 UTC effects: improve `:effect_free`-ness analysis for local mutability (#46200) This commit improves the accuracy of the `:effect-free`-ness analysis, that currently doesn't handle `setfield!` call on local mutable object pretty well. The existing analysis taints `:effect_free`-ness upon any `setfield!` call on mutable object because we really don't have a knowledge about the object lifetime and so we need to conservatively take into account a possibility of the mutable object being a global variable. However we can "recover" `:effect_free`-cness tainted by `setfield!` on mutable object when the newly added `:noglobal` helper effect has been proven because in that case we can conclude that all mutable objects accessed within the method are purely local and `setfield!` on them are `:effect_free` (more precisely we also need to confirm that all the call arguments are known not to be mutable global objects to derive this conclusion). For example now we can prove `:effect_free`-ness of the function below and it will be DCE-eligible (and it will even be concrete-evaluated after #46184): ```julia julia> makeref() = Ref{Any}() makeref (generic function with 1 method) julia> setref!(ref, @nospecialize v) = ref[] = v setref! (generic function with 1 method) julia> @noinline function mutable_effect_free(v) x = makeref() setref!(x, v) x end mutable_effect_free (generic function with 1 method) julia> Base.infer_effects(mutable_effect_free, (String,)) (!c,+e,+n,+t,+s,+g) julia> code_typed() do mutable_effect_free("foo") # will be DCE-ed nothing end 1-element Vector{Any}: CodeInfo( 1 ─ return Main.nothing ) => Nothing ``` 04 August 2022, 05:04:27 UTC
83f8bed Oscar Smith 03 August 2022, 23:07:33 UTC make a bunch of math foldable (#46131) 03 August 2022, 23:07:33 UTC
8dc2f49 pchintalapudi 03 August 2022, 21:29:00 UTC Fix a few JITLink nits (#46216) 03 August 2022, 21:29:00 UTC
df84643 Lilith Orion Hafner 03 August 2022, 13:41:32 UTC make comment load-bearing in test (#46223) Co-authored-by: Lilith Hafner <Lilith.Hafner@gmail.com> 03 August 2022, 13:41:32 UTC
9003732 TheCedarPrince 03 August 2022, 13:39:52 UTC [Docs] Added Example for Using `abspath` (#46218) * Added example and references to abspath docstring 03 August 2022, 13:39:52 UTC
ccc4558 Shuhei Kadowaki 03 August 2022, 10:25:52 UTC effects: improve `:consistent`-cy analysis on `getfield` (#46199) This commit improves the accuracy of the `:consistent`-cy effect analysis by handling `getfield` accessing local mutable objects. The existing analysis taints `:consistent`-cy upon any `getfield` call accessing mutable object because we really don't have a knowledge about the object lifetime and so we need to conservatively take into account a possibility of the mutable object being a global variable. However we can "recover" `:consistent`-cy tainted by `getfield` on mutable object when the newly added `:noglobal` helper effect has been proven because in that case we can conclude that all mutable objects accessed within the method are purely local and thus `:consistent` (more precisely we also need to confirm that all the call arguments are known not to be mutable global objects to derive this conclusion). For example now we can prove `:consistent`-cy of the function below and it will be concrete-evaluated: ```julia julia> @noinline function mutable_consistent(s) broadcast(identity, Ref(s)) end mutable_consistent (generic function with 1 method) julia> Base.infer_effects(mutable_consistent, (String,)) (+c,+e,!n,+t,+s,+g) julia> code_typed() do mutable_consistent(:foo) end 1-element Vector{Any}: CodeInfo( 1 ─ return :foo ) => Symbol ``` 03 August 2022, 10:25:52 UTC
c094a89 Shuhei Kadowaki 03 August 2022, 05:32:59 UTC effects: add `:inaccessiblememonly` effect (#46198) This is a preparatory PR for future improvements on the effects analysis. This commit adds the `:inaccessiblememonly` helper effect, that tracks if a method involves any access or modification on any mutable state. This effect property is basically same as LLVM's `inaccessiblememonly` function attribute, except that it only reasons about mutable memory. This effect property can be considered as a very limited and coarse version of escape analysis and allow us prove `:consistent`-cy or `:effect_free`-ness even in a presence of accesses or modifications on local mutable allocations in cases when we can prove they are local and don't escape. Separate PRs that actually improve the effect analysis accuracy will follow. 03 August 2022, 05:32:59 UTC
eedf3f1 Jameson Nash 02 August 2022, 15:30:58 UTC codegen: truncate Float16 vector ops also (#46130) fix #45881 02 August 2022, 15:30:58 UTC
0b9eda1 Steven G. Johnson 01 August 2022, 14:27:01 UTC define rank(::AbstractVector) (#46169) 01 August 2022, 14:27:01 UTC
d44ba87 Lilith Orion Hafner 01 August 2022, 14:24:16 UTC add reference to keepat! from deleteat! (#46221) 01 August 2022, 14:24:16 UTC
35ac6e1 Dilum Aluthge 01 August 2022, 00:32:37 UTC Add `@testset let...` to the list of methods at the end of the docstring for `@testset` (#46188) Co-authored-by: Keno Fischer <keno@juliacomputing.com> 01 August 2022, 00:32:37 UTC
dbae5a5 TheCedarPrince 30 July 2022, 22:09:39 UTC Link to deepcopy in copy docstring (#46219) 30 July 2022, 22:09:39 UTC
ac44d64 Markus Kuhn 30 July 2022, 20:37:24 UTC How to pass options to julia using #!/usr/bin/env -S (#46205) Option `env -S` for splitting command-line arguments in shebang scripts has been supported on Linux now for a couple of years already, and on FreeBSD and macOS long before that. Therefore we can now recommend using it, and replace the more complicated bash-based method in the FAQ. 30 July 2022, 20:37:24 UTC
acfdc88 Shuhei Kadowaki 30 July 2022, 04:30:33 UTC effects: redesign the `Effects` data structure (#46180) This commit stops representing each effect property as `TriState` but represents them as `UInt8` or `Bool` directly. The motivation is that the tri-state representation hasn't been used actually and rather the incoming improvements on the analysis want to represent some effects as bits to propagate more information. For example, `:consistent`-cy is represented as `UInt8`, where currently it has the following meanings: - `const ALWAYS_TRUE = 0x00` - `const ALWAYS_FALSE = 0x01` - `const CONSISTENT_IF_NOTRETURNED = 0x02`: the `:consistent`-cy can be refined using the return type information later in a case when allocated mutable objects are never returned and I'm also planning to add `const CONSISTENT_IF_NOGLOBAL = 0x04`, that allows us to improve the analysis accuracy by refining the `:consistent`-cy using new effect property that tracks escapability of mutable objects (and actually the similar improvement can be added for `:effect_free`-ness by changing its type from `Bool` to `UInt8` as like `:consistent`-cy). 30 July 2022, 04:30:33 UTC
2e0b75c pchintalapudi 30 July 2022, 04:11:45 UTC Create the new pass manager pipelines (#46175) * Create basic NewPM structures * Replace incidental uses of the legacy pass manager with the new pass manager * Run the MC emitter 30 July 2022, 04:11:45 UTC
6b51780 Jameson Nash 30 July 2022, 00:01:39 UTC dump: ensure we generate the whole backedge graph (#46171) The edge-restore algorithm here is pretty bad now, but this should hopefully fix #45444 30 July 2022, 00:01:39 UTC
e2a8a4e Lilith Orion Hafner 29 July 2022, 11:04:11 UTC Use applicable correctly in sort! (#46194) (also replace hi-lo+1 with len) 29 July 2022, 11:04:11 UTC
8fe9d93 Ian Atol 29 July 2022, 10:22:03 UTC Support begin and end in gen_call_with_extracted_types users (#45968) 29 July 2022, 10:22:03 UTC
823766e Shuhei Kadowaki 29 July 2022, 10:18:47 UTC inference: fix #46207, make sure we never form nested `Conditional` (#46208) 29 July 2022, 10:18:47 UTC
255acea DilumAluthgeBot 29 July 2022, 03:56:33 UTC 🤖 Bump the Pkg stdlib from 7920ff4d5 to cebcbc0a4 (#46206) Co-authored-by: Dilum Aluthge <dilum@aluthge.com> 29 July 2022, 03:56:33 UTC
6f737f1 Keno Fischer 28 July 2022, 14:07:30 UTC Print more context when FileWatching test fails; also make small tweaks to the `ContextTestSet` (`@testset let ...`) functionality in the Test stdlib (#46186) Co-authored-by: Dilum Aluthge <dilum@aluthge.com> 28 July 2022, 14:07:30 UTC
9630911 Jeff Bezanson 27 July 2022, 22:14:53 UTC mark `setfield!` as consistent (#46184) 27 July 2022, 22:14:53 UTC
b2bf56e Lilith Orion Hafner 27 July 2022, 15:43:14 UTC Throw descriptive BoundsError in copyto! (#46192) 27 July 2022, 15:43:14 UTC
dd2095a Dilum Aluthge 27 July 2022, 06:43:35 UTC Fix a typo in the docstring for `@testset let` in the Test stdlib (#46187) 27 July 2022, 06:43:35 UTC
60d1af9 Shuhei Kadowaki 26 July 2022, 23:19:17 UTC effects: flip the meaning of `inbounds_taint_consistency` (#46179) This commit renames `inbounds_taint_consistency` to `noinbounds` so that the meaning is flipped. This should be more aligned with the other effects, where they generally mean an absence of "undesirable" properties, e.g. `nonoverlayed`. 26 July 2022, 23:19:17 UTC
a351d25 woclass 26 July 2022, 21:36:14 UTC devdoc/build: remove unused step (#46125) 26 July 2022, 21:36:14 UTC
ff1b563 Keno Fischer 26 July 2022, 21:08:20 UTC Print more context for complex test conditions (#46138) On some failures, e.g. in the cmdlineargs failure in [1], printing the failing expression doesn't give us very much information, because what we're doing is just testing a small part of a larger structure, but to really diagnose the failure, we'd really want to see the whole structure (in this case the stderr output, not just the failure bool). This introduces an additional `@testset` feature that lets failures optionally include a context variable that gets printed on failure. Example: ``` julia> @testset let v=(1,2,3) @test v[1] == 1 @test v[2] == 3 end Test Failed at REPL[8]:3 Expression: v[2] == 3 Evaluated: 2 == 3 Context: v = (1, 2, 3) ERROR: There was an error during testing ``` The syntax here is `@testset let`, which was previously unused. The testset is transparent and failures/successes are passed through directly to the parent testset. In particular, using this kind of testset does not cause additional nesting in the default testset print. [1] https://buildkite.com/julialang/julia-master/builds/14160#01822311-84f0-467a-a8af-a5d751f2c6ab/448-728 26 July 2022, 21:08:20 UTC
5d2e24f Ian Atol 26 July 2022, 19:58:30 UTC Change inlineable field to store inlining cost (#45378) 26 July 2022, 19:58:30 UTC
9e22e56 Shuhei Kadowaki 26 July 2022, 14:21:47 UTC tests: allow `test/compiler/AbstarctInterpreter.jl` to be loaded multiple times (#46172) It seems better to make it able to load that file multiple times rather than trying hard to not pollute the namespace. 26 July 2022, 14:21:47 UTC
2224789 Daniel Karrasch 26 July 2022, 08:48:45 UTC Save matrix allocation in `Matrix(::AbstractQ)` (#46162) 26 July 2022, 08:48:45 UTC
0ea2b2d N5N3 25 July 2022, 22:34:03 UTC `copyto!` fix for `BitArray`/`AbstractArray`, fixes #25968 (#46161) 1. map `copyto!(::BitArray, n1, ::BitArray, n2, l)` to `Base.unsafe_copyto!` 2. add missing unaliasing in `copyto!` for `AbstractArray` 25 July 2022, 22:34:03 UTC
1addb84 Keno Fischer 25 July 2022, 19:45:28 UTC mac: Produce coredumps on segfault (#46157) This changes the mach exception server to ignore fatal SIGSEGVs, letting regular kernel processing handle it (by performing POSIX signal delivery and then subsequently coredumping), rather than quitting the process directly. There's probably some way to induce the kernel to perform core dumping directly from the exception server, but I think it'll be less confusing all around to just have segfaults take the standard path. Hoping this will help debug #46152. 25 July 2022, 19:45:28 UTC
2982986 Shuhei Kadowaki 25 July 2022, 19:39:12 UTC effects: refactor `builtin_effects` (#46097) Factor special builtin handlings into separated functions (e.g. `getfield_effects`) so that we can refactor them more easily. This also improves analysis accuracy a bit, e.g. ```julia julia> Base.infer_effects((Bool,)) do c obj = c ? Some{String}("foo") : Some{Symbol}(:bar) return getfield(obj, :value) end (!c,+e,!n,+t,!s) # master (+c,+e,!n,+t,+s) # this PR ``` 25 July 2022, 19:39:12 UTC
73c1eeb Oscar Smith 25 July 2022, 19:21:43 UTC fix rem2pi for non-finite arguments (#46163) 25 July 2022, 19:21:43 UTC
69e319d Jameson Nash 25 July 2022, 18:26:34 UTC tcp: re-enable half-duplex operation support (#46088) Refs: #42005 25 July 2022, 18:26:34 UTC
8a8e3bf Shuhei Kadowaki 25 July 2022, 14:20:34 UTC Merge pull request #46111 from JuliaLang/avi/correct-consistent effects: fix correctness issues of `:consistent`-cy analysis 25 July 2022, 14:20:34 UTC
783a6aa Dilum Aluthge 25 July 2022, 01:15:53 UTC Revert broadening of arithmetic `Any` methods (revert #44564 and #45320) (#45489) * Revert "Remove type-unlimited unary `+` and `*` (#45320)" This reverts commit 990b1f3b1a254963bd71ceada44a560161225afb. * Revert "Generalize or restrict a few basic operators (#44564)" This reverts commit cf1f717700ada2784575c969d7114b3416e4f138. Also fixes merge conflicts in stdlib/LinearAlgebra/test/generic.jl Co-authored-by: Daniel Karrasch <daniel.karrasch@posteo.de> 25 July 2022, 01:15:53 UTC
665b03e Gautam Mishra 24 July 2022, 22:16:01 UTC Fix rem2pi for NaN inputs, fixes #32888. (#36420) 24 July 2022, 22:16:01 UTC
e3de4a8 Sukera 24 July 2022, 18:28:15 UTC Add Metaprogramming manual reference to `macro` docstring (#46079) Co-authored-by: Shuhei Kadowaki <40514306+aviatesk@users.noreply.github.com> 24 July 2022, 18:28:15 UTC
54881e2 Shuhei Kadowaki 20 July 2022, 02:53:20 UTC effects: taint `:consistent`-cy on allocation/access of uninitialized fields 24 July 2022, 18:21:46 UTC
6e1ded4 Shuhei Kadowaki 20 July 2022, 00:15:58 UTC effects: taint `:consistent`-cy on `:the_exception` expression 24 July 2022, 18:16:08 UTC
cee90db Lilith Orion Hafner 23 July 2022, 21:39:54 UTC simplify/style in Dates.CompoundPeriod (#46037) 23 July 2022, 21:39:54 UTC
8f54c41 Lilith Orion Hafner 23 July 2022, 21:24:15 UTC Test that sorting preserves object identity (#46102) To make sure that #39620 stays closed 23 July 2022, 21:24:15 UTC
fb760d9 Keno Fischer 23 July 2022, 19:57:50 UTC In IR2OC constructor, widenconst OC argtypes (#46108) * In IR2OC constructor, widenconst OC argtypes The IR argtypes are lattice elements, but OC needs types. * Update base/compiler/ssair/legacy.jl Co-authored-by: Shuhei Kadowaki <40514306+aviatesk@users.noreply.github.com> Co-authored-by: Shuhei Kadowaki <40514306+aviatesk@users.noreply.github.com> 23 July 2022, 19:57:50 UTC
2876f53 Nicholas Bauer 23 July 2022, 16:22:59 UTC Add tests for hvncat to avoid issues with non-numeric types (#46145) 23 July 2022, 16:22:59 UTC
f1991ed Shuhei Kadowaki 23 July 2022, 00:37:38 UTC export `[@]invokelatest` (#45831) 23 July 2022, 00:37:38 UTC
c32c0d2 Shuhei Kadowaki 23 July 2022, 00:16:48 UTC remove dead comment (#46132) #45993 fixed this bug, let's make the math effect-tests actually fail 23 July 2022, 00:16:48 UTC
c5a63d8 Lionel Zoubritzky 22 July 2022, 20:45:03 UTC Add REPL-completions for keyword arguments (#43536) 22 July 2022, 20:45:03 UTC
9bdaabd woclass 22 July 2022, 18:15:06 UTC Makefile: fix `PATH` set on check-whitespace (#45957) 22 July 2022, 18:15:06 UTC
f3eb156 Keno Fischer 22 July 2022, 17:19:33 UTC Fix opaque closure inlining performance (#46136) At some point over the last few months, the opaque closure inlining path got an extra check for `isdispatchtuple` that is both unnecessary and causes significant performance regressions for Diffractor. Get rid of it and add a test to make sure this doesn't regress again. 22 July 2022, 17:19:33 UTC
6ce65d7 Keno Fischer 22 July 2022, 16:11:11 UTC Strengthen assume_effects admonition (#46134) I don't think people are quite as scared of `@assume_effects` as they should be yet, so try to make the warning slightly more scary and in particular point out that the set of asserted effects should be minimized. 22 July 2022, 16:11:11 UTC
a1d8b10 Oscar Smith 22 July 2022, 12:53:50 UTC 2 arg getfield is always consistent (#46044) * 2 arg getfield is always consistent * make Float64^Float64 foldable and add tests Co-authored-by: Shuhei Kadowaki <aviatesk@gmail.com> 22 July 2022, 12:53:50 UTC
bde1c71 Oscar Smith 22 July 2022, 12:53:13 UTC Better getfield error message (#46129) * add error message for getfield with non-Int Integer types 22 July 2022, 12:53:13 UTC
743578a N5N3 22 July 2022, 05:11:25 UTC Fix effect override for `@ccall` (#46135) 22 July 2022, 05:11:25 UTC
6009ae9 Mosè Giordano 21 July 2022, 23:21:33 UTC [LinearAlgebra] Support more env variables to set OpenBLAS threads (#46118) 21 July 2022, 23:21:33 UTC
d75843d Gabriel Baraldi 21 July 2022, 19:52:21 UTC Correct codegen bugs introduced by allocation-hoisting-PR (#45476) Co-authored-by: Jameson Nash <vtjnash@gmail.com> 21 July 2022, 19:52:21 UTC
46a6f22 DilumAluthgeBot 21 July 2022, 16:15:01 UTC 🤖 Bump the Downloads stdlib from c34ec3e to 0733701 (#46107) Co-authored-by: Dilum Aluthge <dilum@aluthge.com> 21 July 2022, 16:15:01 UTC
e163f5a Markus Kuhn 21 July 2022, 15:51:45 UTC 1.8 NEWS item: Distributed SSHManager csh support (#46126) 21 July 2022, 15:51:45 UTC
a4b91df Shuhei Kadowaki 21 July 2022, 14:59:53 UTC effects: mark `svec` as `:consistent` (#46120) 21 July 2022, 14:59:53 UTC
799a2cf Shuhei Kadowaki 21 July 2022, 14:49:23 UTC inference: simplify error-case handling within `abstract_eval_statement` (#46119) 21 July 2022, 14:49:23 UTC
76d56ff Shuhei Kadowaki 20 July 2022, 23:08:32 UTC inference: form `PartialStruct` for mutable objects with `const` field declaration (#46117) By exploiting the semantic constraint imposed by new `const` declaration we can sometimes form `PartialStruct` for mutable objects (when there is some information available to refine the type of `const`-declared field) to allow extra type information propagation. 20 July 2022, 23:08:32 UTC
f3c2227 Ian Butterworth 20 July 2022, 15:51:39 UTC Fix timev compilation time tracking and add tests (#46100) 20 July 2022, 15:51:39 UTC
028e9ff Shuhei Kadowaki 20 July 2022, 15:15:49 UTC fix `_builtin_nothrow` for `arrayset` (#46105) 20 July 2022, 15:15:49 UTC
0c03238 Julian Samaroo 20 July 2022, 14:43:25 UTC Explicitly derive alloca AS from DL (#45900) * Explicitly derive alloca AS from DL Fixes issues around usage of the AMDGPU LLVM target by GPUCompiler * Use alloca AS for sret pointer type * Insert addrspacecast in stringConstPtr * text/llvmpasses: Test non-0 alloca addrspace Co-authored-by: Collin Warner <collinw@mit.edu> 20 July 2022, 14:43:25 UTC
6a41ec1 DilumAluthgeBot 20 July 2022, 09:08:39 UTC 🤖 Bump the Pkg stdlib from 2beb40c42 to 7920ff4d5 (#46096) Co-authored-by: Dilum Aluthge <dilum@aluthge.com> 20 July 2022, 09:08:39 UTC
97df6db Chris Elrod 20 July 2022, 08:16:13 UTC Use `max(1, Sys.CPU_THREADS)` BLAS threads for apple `aarch64`. (#46085) 20 July 2022, 08:16:13 UTC
d7c56ba Tim Besard 20 July 2022, 08:08:27 UTC Specialize tuple setindex to avoid ntuple-related performance regression. (#46050) 20 July 2022, 08:08:27 UTC
017228a N5N3 20 July 2022, 08:04:46 UTC Fix fast path for `strides(::ReinterpretArray)` with dense parent. (#46114) 20 July 2022, 08:04:46 UTC
8378d4c Valentin Churavy 20 July 2022, 02:54:15 UTC Merge pull request #45890 from fingolfin/mh/export-jl_is_foreign_type Export jl_is_foreign_type 20 July 2022, 02:54:15 UTC
adf2e1b Valentin Churavy 20 July 2022, 00:16:20 UTC Merge pull request #45649 from JuliaLang/vc/fp16 Emit aliases to FP16 conversion routines 20 July 2022, 00:16:20 UTC
09c587c Keno Fischer 19 July 2022, 21:45:31 UTC Relax type of `PartialOpaque` source field (#46087) We type this as `::Method`, which is what's expected in Base. However, Diffractor wants to be able to generate the source lazyily, so it needs to be able to stash some information here to be able to perform type inference before generating the method source. When that code was originally written (pre-OpaqueClosure), this used a custom lattice element, but `PartialOpaque` seems appropriate, since these do widen to an OpaqueClosure and what's happening here is essentially just that the `source` is being generated lazily. We may want to revisit this if we switch the way that we represent lattice elements, but for now, this seems like the path of least resistance. 19 July 2022, 21:45:31 UTC
57ffed6 Valentin Churavy 19 July 2022, 18:33:35 UTC Merge pull request #46089 from rikhuijzer/patch-1 Mention `PProf` in list of profile browsers 19 July 2022, 18:33:35 UTC
808ad85 Lilith Orion Hafner 19 July 2022, 17:21:01 UTC Fix sort(Union{}[]) (#46095) 19 July 2022, 17:21:01 UTC
3407fb3 Valentin Churavy 12 June 2022, 15:59:05 UTC Emit aliases into the system image - Put the interposer in llvm.compiler.used. - Injecting the aliases after optimization: Our multiversioning pass interacts badly with the llvm.compiler.used gvar. Co-authored-by: Tim Besard <tim.besard@gmail.com> Co-authored-by: Valentin Churavy <v.churavy@gmail.com> 19 July 2022, 16:43:33 UTC
ff36015 Valentin Churavy 11 June 2022, 21:26:34 UTC Define aliases to FP16 crt in the OJIT 19 July 2022, 16:43:33 UTC
f651866 Jameson Nash 11 June 2022, 20:59:49 UTC Prefix Float16 intrinsics 19 July 2022, 16:43:33 UTC
db570df Lilith Orion Hafner 19 July 2022, 15:05:30 UTC Stop using view in adaptive sort (#45699) 19 July 2022, 15:05:30 UTC
9863085 Rik Huijzer 19 July 2022, 07:56:42 UTC Mention `PProf` in list of profile browsers 19 July 2022, 07:56:42 UTC
82c3a6f woclass 18 July 2022, 20:39:52 UTC Fix codegen test on debug version (#46065) * test: fix codegen debug lib name * test: skip some codegen tests in debug version 18 July 2022, 20:39:52 UTC
820c08b Jeff Bezanson 18 July 2022, 19:02:46 UTC fix #45825, BitArray methods assuming 1-indexing of AbstractArray (#45835) 18 July 2022, 19:02:46 UTC
c01a1e2 Valentin Churavy 18 July 2022, 12:46:31 UTC Merge branch 'master' into mh/export-jl_is_foreign_type 18 July 2022, 12:46:31 UTC
e1739aa t-bltg 18 July 2022, 10:27:38 UTC fix `typejoin` docstring (#46018) Co-authored-by: Shuhei Kadowaki <40514306+aviatesk@users.noreply.github.com> 18 July 2022, 10:27:38 UTC
94e4082 Jameson Nash 18 July 2022, 09:51:17 UTC make convert(Union{},x) directly ambiguous (#46000) This should make it impossible to accidentally define or call this method on foreign types. Refs: #31602 Fixes: #45837 Closes: #45051 18 July 2022, 09:51:17 UTC
d8a90e4 DilumAluthgeBot 18 July 2022, 09:49:28 UTC 🤖 Bump the Pkg stdlib from 56cd041cc to 2beb40c42 (#46057) Co-authored-by: Dilum Aluthge <dilum@aluthge.com> 18 July 2022, 09:49:28 UTC
6b91a82 Jeff Bezanson 18 July 2022, 09:44:16 UTC fix #46051, OOB string index in printing irrationals (#46056) 18 July 2022, 09:44:16 UTC
29586ef Ian Butterworth 17 July 2022, 20:12:51 UTC remove examples of time_imports nesting given it was a bug (#46072) 17 July 2022, 20:12:51 UTC
425d5e0 Valentin Churavy 17 July 2022, 18:53:58 UTC Merge pull request #46063 from JuliaLang/tb/llvm Update LLVM to include additional patches. 17 July 2022, 18:53:58 UTC
707f59b Mosè Giordano 17 July 2022, 18:14:07 UTC [nghttp2_jll] Upgrade to v1.48.0 (#45980) * [nghttp2_jll] Upgrade to v1.48.0 * [nghttp2_jll] Fix test 17 July 2022, 18:14:07 UTC
3fd61e4 Kristoffer Carlsson 17 July 2022, 06:03:07 UTC document and export `samefile` (#45275) Co-authored-by: Viral B. Shah <ViralBShah@users.noreply.github.com> Co-authored-by: Denis Barucic <barucden@fel.cvut.cz> 17 July 2022, 06:03:07 UTC
44e9b87 Lilith Orion Hafner 17 July 2022, 04:15:14 UTC Fix semantic error in effect documentation (#46042) Co-authored-by: Lilith Hafner <Lilith.Hafner@gmail.com> 17 July 2022, 04:15:14 UTC
7261c65 Tim Besard 16 July 2022, 13:11:41 UTC Show IR verification failures. (#46062) 16 July 2022, 13:11:41 UTC
back to top