affe96a | Milan Bouchet-Valat | 17 January 2021, 17:30:37 UTC | Simplify computation of return type in broadcast Since we rely on inference, we can use `_return_type` directly instead of via a complex machinery. | 17 January 2021, 17:32:12 UTC |
c487dd0 | Jameson Nash | 10 January 2021, 19:28:43 UTC | inference: fix vararg normalization in rewrap (#39134) Fixes #39082 | 10 January 2021, 19:28:43 UTC |
9e7e23d | Valentin Churavy | 10 January 2021, 17:24:39 UTC | Merge pull request #39166 from Thalassocracy/fix-documentation Replace broken link in distributing.md | 10 January 2021, 17:24:39 UTC |
927c9fa | Kenta Sato | 10 January 2021, 13:04:39 UTC | make lowrankupdate/downdate more permissive (#39150) * make lowrankupdate/downdate more permissive * add tests for abstract vectors | 10 January 2021, 13:04:39 UTC |
aea60ed | Thalassocracy | 10 January 2021, 00:50:50 UTC | Replace broken link in distributing.md | 10 January 2021, 00:50:50 UTC |
ffa966e | Mark Kittisopikul | 09 January 2021, 23:49:00 UTC | Remove start_repl_server code from REPL module (#39162) `REPL.start_repl_server(port::Int)` uses outdated API and no longer functions. It should be removed. ```julia julia> import REPL julia> REPL.start_repl_server(8080) ERROR: MethodError: no method matching listen(::REPL.var"#80#81", ::Int64) ``` | 09 January 2021, 23:49:00 UTC |
e57501c | Shuhei Kadowaki | 09 January 2021, 23:43:31 UTC | minor code quality improvements (#39159) | 09 January 2021, 23:43:31 UTC |
1393310 | Marius Millea | 09 January 2021, 20:02:56 UTC | fix logdet(Diagonal{<:Real}) with negative entries (#39158) | 09 January 2021, 20:02:56 UTC |
8937f7e | Jameson Nash | 09 January 2021, 04:17:39 UTC | gf: avoid adding cache entries wider than the original method (#39140) Sometimes we want to widen the compilation signature, but then end up with something which does not fit the original pattern. This then can cause problems later, when we try to use the Method (from the cache), but discover it does not actually match the call. Fixes #38999 | 09 January 2021, 04:17:39 UTC |
ca07546 | Simeon Schaub | 08 January 2021, 18:34:09 UTC | fix count(::BitArray; dims) (#39149) Not sure why this worked before #37461, perhaps #9498? | 08 January 2021, 18:34:09 UTC |
438f9cb | Jameson Nash | 08 January 2021, 18:32:56 UTC | Core printing: improve show of methods (#39135) This adds their environment, so that their TypeVars will print correctly. | 08 January 2021, 18:32:56 UTC |
6cff73f | Jeff Bezanson | 08 January 2021, 09:32:34 UTC | make methods of `alloc_buf_hook` return the same types (#39141) | 08 January 2021, 09:32:34 UTC |
83bee67 | Jeff Bezanson | 08 January 2021, 02:28:41 UTC | fix some compiler warnings (#39142) - unused jl_iterate_func - cast type of realloc in jl_init_threading | 08 January 2021, 02:28:41 UTC |
6c42190 | Debaditya Pal | 07 January 2021, 20:27:24 UTC | added automatic keyword assignment support to test macro (#38270) * added automatic keyword assignment support to @test macro * added some tests for test macro using atol keyword * x = a.x syntax support added * Update stdlib/Test/src/Test.jl Co-authored-by: Simeon Schaub <simeondavidschaub99@gmail.com> Co-authored-by: Simeon Schaub <simeondavidschaub99@gmail.com> | 07 January 2021, 20:27:24 UTC |
34ab0a9 | Valentin Churavy | 07 January 2021, 18:56:43 UTC | Merge pull request #39051 from JuliaLang/vc/sext Annotate function arguments with sext/zext | 07 January 2021, 18:56:43 UTC |
66c9f6a | Jeff Bezanson | 07 January 2021, 16:58:28 UTC | fix #39117, overly-specific type from `default_eltype` (#39130) | 07 January 2021, 16:58:28 UTC |
59eb9f9 | Jameson Nash | 07 January 2021, 16:57:16 UTC | remove Core._apply Builtin (#39115) We can emulate this deprecated function, until we delete it in v2. This in turn revealed some missing error checks, which we also add. Fixes #39113 | 07 January 2021, 16:57:16 UTC |
fa14d0b | Elliot Saba | 07 January 2021, 16:55:50 UTC | `addenv()`, by default, should inherit from `ENV` (#39100) The spirit of `addenv()` is to be non-destructive, however because a `Cmd` object with an `env` member that is set to `nothing` (the default) inherts from the current environment when it is run. This means that the following, rather surprising behavior holds true on current master: ``` julia> ENV["FOO"] = "foo" run(`/bin/bash -c "echo \$FOO \$BAR"`) run(addenv(`/bin/bash -c "echo \$FOO \$BAR"`, "BAR" => "bar")) foo bar ``` This PR adds an `inherit` flag to `addenv()` to allow keeping this behavior if it is actually desired (this might make sense if you are constructing `.env` blocks over a series of `addenv()` calls and you don't want to have to special-case your first operation as a `setenv()` rather than an `addenv()`. | 07 January 2021, 16:55:50 UTC |
c70a5bc | Jeff Bezanson | 07 January 2021, 16:51:48 UTC | fix part of #38936, getfield elim handling union of tuples (#39107) This fixes the regression since 1.5 | 07 January 2021, 16:51:48 UTC |
f5c8e67 | Tim Besard | 07 January 2021, 16:49:43 UTC | Fix alloc opt pass on LLVM 11+. (#39119) | 07 January 2021, 16:49:43 UTC |
a319ae4 | Jameson Nash | 07 January 2021, 15:45:27 UTC | Revert "Improve copyto! for short heterogeneous tuples (#39035)" (#39128) This reverts commit b00e9f0bac05de913d5b869739bcc93ac259c9f7, which has a 1/64 chance of failing tests. | 07 January 2021, 15:45:27 UTC |
03957db | Ian Butterworth | 07 January 2021, 15:36:15 UTC | Remove try from at-time and close compile timer during throw (#39133) * remove try from at-time and close compile timer during throw * add scope tests for at-time and aat-timev | 07 January 2021, 15:36:15 UTC |
86639e1 | Jose Manuel de Frutos | 07 January 2021, 06:20:18 UTC | Add a warning that HTML should not be used #38909 (#39072) * Add a warning that HTML should not be used #38909 | 07 January 2021, 06:20:18 UTC |
0949eb2 | Valentin Churavy | 30 December 2020, 19:50:22 UTC | silence unused variable warning | 06 January 2021, 19:45:22 UTC |
5c6e21e | Shuhei Kadowaki | 06 January 2021, 18:22:03 UTC | inference: minor typing improvements for `InferenceState.stmt_types` (#39096) | 06 January 2021, 18:22:03 UTC |
ed89ae7 | Mustafa M | 06 January 2021, 17:00:37 UTC | Test for Windows delete permissions in system folders (#39078) | 06 January 2021, 17:00:37 UTC |
9f8bdc2 | Simeon Schaub | 06 January 2021, 08:09:40 UTC | fix Meta.partially_inline! again (#39112) forgot to take `statement_offset` into account in #38519 | 06 January 2021, 08:09:40 UTC |
158e22f | Ian Butterworth | 06 January 2021, 08:09:19 UTC | Measure compile time only when using time macros pt.2: handling when code under test throws and make compile timing thread-local (#38915) * ensure compile timing disables in time & timev macros * make comp measurement switching threadsafe | 06 January 2021, 08:09:19 UTC |
f46da9e | Valentin Churavy | 30 December 2020, 19:45:48 UTC | mark integer parameters as sext/zext | 06 January 2021, 00:22:09 UTC |
399f8ba | Keno Fischer | 05 January 2021, 22:27:17 UTC | Merge pull request #39101 from JuliaLang/sf/sandbox_fixes Fixes so that test suite can run in more tightly sandboxed environment | 05 January 2021, 22:27:17 UTC |
3931b22 | Mustafa M | 05 January 2021, 18:15:52 UTC | filesystem: use libuv for rmdir directly (#39013) | 05 January 2021, 18:15:52 UTC |
ee6f6f1 | mroavi | 05 January 2021, 09:50:03 UTC | fix typo in metaprogramming.md (#39095) Fix typo | 05 January 2021, 09:50:03 UTC |
fd671b3 | Elliot Saba | 05 January 2021, 08:35:46 UTC | [REPL] Do not require test to run within `/tmp` | 05 January 2021, 08:35:46 UTC |
7a6f423 | Elliot Saba | 05 January 2021, 07:59:57 UTC | [REPL]: Don't write to the user's home directory Instead of writing out to actual `$HOME`, create a temporary directory and set that as `$HOME`/`$USERPROFILE`. | 05 January 2021, 07:59:57 UTC |
383db92 | Morten Piibeleht | 05 January 2021, 07:44:11 UTC | Bump Documenter to 0.26.1 (#39097) | 05 January 2021, 07:44:11 UTC |
7dae29d | Elliot Saba | 05 January 2021, 06:48:34 UTC | [Artifacts]: Download to temporary directories, not `stdlib` We can't assume that `stdlib` is in a writable location; always download temporary testing files to `$TMPDIR`. | 05 January 2021, 06:48:34 UTC |
1a333a9 | Mathieu Besançon | 04 January 2021, 23:44:04 UTC | fix inf rational (#39063) | 04 January 2021, 23:44:04 UTC |
ee66728 | Sacha Verweij | 04 January 2021, 22:59:35 UTC | Merge pull request #39080 from aviatesk/typingedges minor typing improvements for `(frame::InferenceState).stmt_edges` | 04 January 2021, 22:59:35 UTC |
e2f4073 | Sheehan Olver | 04 January 2021, 19:44:04 UTC | Fixes for non-Int based lengths (#37741) | 04 January 2021, 19:44:04 UTC |
1dc3804 | Keno Fischer | 04 January 2021, 18:21:04 UTC | Small drive-by cleanup: Make constant explicit (#39056) This code is a bit of a leftover from when we still had `nfields(::Type)`, but at this point it was always just returning the number of fields of `DataType`. I was a bit confused by this code, so I thought just using `jl_datatype_t` here explicitly makes clear that it does actually just return said constant. | 04 January 2021, 18:21:04 UTC |
974e6ec | Lyndon White | 04 January 2021, 17:32:19 UTC | Remove outdated TODO about string encodings (#39077) | 04 January 2021, 17:32:19 UTC |
8d643a2 | Stefano Sgorlon | 04 January 2021, 04:44:08 UTC | LICENSE: copyright end date 2019 -> 2021 (#39084) | 04 January 2021, 04:44:08 UTC |
628b7ac | Shuhei Kadowaki | 03 January 2021, 03:58:15 UTC | minor typing improvements for `(frame::InferenceState).stmt_edges` | 03 January 2021, 03:58:15 UTC |
5ee2d60 | Shikhar Goswami | 01 January 2021, 10:48:37 UTC | Written tests for typemin/typemax (#39057) * Written tests for typemin/typemax * Added test w.r.t Int128 | 01 January 2021, 10:48:37 UTC |
82d79ce | Keno Fischer | 31 December 2020, 07:00:17 UTC | Add `ismutabletype` (#39037) Determines whether a type was declared using `mutable struct`. Naming follows from the `ismutable` query we already have, analogous to `isbits`/`isbitstype`. Replaces #18168. | 31 December 2020, 07:00:17 UTC |
b00e9f0 | Keno Fischer | 31 December 2020, 06:59:47 UTC | Improve copyto! for short heterogeneous tuples (#39035) There was a complaint on Twitter [1], that Julia is slower than Ruby(!) at allocating short heterogeneous arrays. The code looks something like this: ``` f() = for i in 1:100000; b=[1,2.0]; end; ``` Of course, this benchmark is silly because the arrays are unused, so the entire code should be dropped once we have further improvements to our lifetime analysis, which will change the runtime 0. However, since we're not quite able to do that yet, it is somewhat unexpected that we should be slower than Ruby here. Admittedly our arrays are tuned for larger sizes, but still. Upon further investigation, it turns out that the issue is that we're allocating extra boxed objects when pulling the fields out of the literal, since we don't have any specialization for `copyto!` from tuples, falling back to the generic `copyto!` method, which is type unstable in the presence of heterogeneous tuples. Since this code is reachable from surface syntax, I think it's worth putting in a couple of specializations to accelerate the array construction, which is what this PR does (up to tuples of length 10 to avoid excessive codegen). Since the type-instability issue applies whether the value is used or not, this should still be useful even if we're able to prove that the allocations can be dropped entirely. Before: ``` julia> @benchmark f() BenchmarkTools.Trial: memory estimate: 21.36 MiB allocs estimate: 600000 -------------- minimum time: 14.928 ms (0.00% GC) median time: 16.297 ms (0.00% GC) mean time: 16.323 ms (3.84% GC) maximum time: 26.042 ms (0.00% GC) -------------- samples: 307 evals/sample: 1 ``` After: ``` julia> @benchmark f() BenchmarkTools.Trial: memory estimate: 9.16 MiB allocs estimate: 100000 -------------- minimum time: 2.117 ms (0.00% GC) median time: 2.276 ms (0.00% GC) mean time: 2.455 ms (7.62% GC) maximum time: 4.638 ms (28.99% GC) -------------- samples: 2036 evals/sample: 1 ``` Which puts us about on par with Ruby on my machine (not entirely unsurprising, since this is essentially an allocation benchmark at this point and both allocators are written in C). | 31 December 2020, 06:59:47 UTC |
1ad6aed | Keno Fischer | 31 December 2020, 03:54:08 UTC | Merge pull request #39052 from JuliaLang/sf/win_embedding_fixes Various fixes to trampolines for embedding on Windows | 31 December 2020, 03:54:08 UTC |
ed9dde2 | Katharine Hyatt | 31 December 2020, 03:48:21 UTC | Test for checkout_head (#39034) * Test for checkout_head * Remember to actually close the file * And actually force the checkout | 31 December 2020, 03:48:21 UTC |
8981e0b | Elliot Saba | 30 December 2020, 22:27:00 UTC | [cli/loader]: Don't allow initialization to run more than once There are multiple ways of fixing this, but let's just not let trampolines get re-set after they've already been set. | 30 December 2020, 22:27:00 UTC |
b26a386 | Elliot Saba | 30 December 2020, 22:25:32 UTC | [cli/trampolines] Fix section directive for windows trampolines By naming the section `text` instead of `.text` these sections were mapped into memory without the executable bit set, causing segfaults when attempting to use any trampolines on Windows. | 30 December 2020, 22:25:32 UTC |
144f781 | Elliot Saba | 30 December 2020, 22:24:12 UTC | [loader/libjulia]: Call `setup_stdio()` when loading `libjulia` This allows libjulia's initialization routines to properly output to `stdout` in embedding scenarios. | 30 December 2020, 22:24:12 UTC |
3d922e4 | Simeon Schaub | 30 December 2020, 19:00:17 UTC | specialize on mapfoldl(::Type, ...) (#39019) Before: ```julia julia> using BenchmarkTools julia> b = rand(Bool, 10000); julia> @btime mapfoldl(Int, +, b; init=0) 1.136 ms (9025 allocations: 141.02 KiB) 5057 ``` After: ```julia julia> @btime mapfoldl(Int, +, b; init=0) 1.094 μs (1 allocation: 16 bytes) 5057 ``` | 30 December 2020, 19:00:17 UTC |
31eb6c2 | Sebastian Stock | 30 December 2020, 18:03:27 UTC | Avoid StepRangeLen eltype widening (#38979) | 30 December 2020, 18:03:27 UTC |
d46198d | Elliot Saba | 30 December 2020, 07:26:33 UTC | [libuv] Bump to include `DELETE` win ACL patch (#39038) This bump to LibUV adds an important permission to files which should allow them to be deleted more reliably when created in "system"-like locations such as `C:\`. | 30 December 2020, 07:26:33 UTC |
aeb933c | Keno Fischer | 30 December 2020, 04:57:00 UTC | Merge pull request #39036 from JuliaLang/kf/compilerroundup Round up and rebase approved compiler PRs | 30 December 2020, 04:57:00 UTC |
8a2ccf2 | Yichao Yu | 06 September 2020, 02:15:02 UTC | Some `getfield` related fixes in inference * Use the field index passed in in `lift_leaves` The caller has already done all the computation including bound checking. The `field` computed in this function is also affecting all the following iterations which is almost certainly wrong. * Remove unnecessary type check on `field` in `lift_leaves` since it is always `Int` * Move a branch disabling `return nothing` higher up * Remove some duplicated calculation on field index in `getfield_elim_pass!` * Fix `try_compute_fieldidx` to return `nothing` for non-`Int` `Integer` field index. This can cause `getfield_nothrow` to return incorrect result. It also gives the caller worse type info about the return value. * Teach `getfield_nothrow` that `isbits` field cannot be undefined and getfield on such field cannot throw. This is already handled in `isdefined_tfunc`. * Fix a few wrong use of `isbits` in dead branches ---- Ref #26948 (fa02d34496eb2bed9a55e4f77cb135a6b4180edc) Ref #27126 (9100329d189b3d2ac6d1e03bfc7431cbe8d18c00) | 30 December 2020, 02:04:13 UTC |
bb2e065 | Yichao Yu | 05 September 2020, 18:19:26 UTC | Fix Dict key check in type inference This has been there since #27126 (9100329d189b3d2ac6d1e03bfc7431cbe8d18c00) Dudging by the use of `reverse_mapping[stmt_val]` below this should either be a branch condition or an assert. | 30 December 2020, 02:03:45 UTC |
ad7e59a | Keno Fischer | 29 December 2020, 21:08:46 UTC | Document destructuring syntax in anonymous function argument (#39024) When destructuring a single tuple in an anonymous function argument, you have to include an extra comma to force a 1-tuple. Otherwise ((x,y)) collapses to (x,y). This is not the case for named functions. Co-authored-by: Graham Smith <grahamas@gmail.com> | 29 December 2020, 21:08:46 UTC |
345ce78 | Kristoffer Carlsson | 29 December 2020, 17:00:58 UTC | remove unnecessary ::Union{} type asserts (#35817) * remove unnecessary ::Union{} type asserts | 29 December 2020, 17:00:58 UTC |
ae3a3cc | Daniel Karrasch | 29 December 2020, 08:43:39 UTC | Improve performance for dot(AdjTrans,AdjTrans) (#39004) * Improve performance for dot(AdjTrans,AdjTrans) * generalize to Adjoint/Transpose | 29 December 2020, 08:43:39 UTC |
fc3fe4f | Jeff Bezanson | 29 December 2020, 07:45:49 UTC | fix IR showing not to show TypedSlot type twice (#38961) * fix IR showing not to show TypedSlot type twice * Update base/compiler/ssair/show.jl Co-authored-by: Jameson Nash <vtjnash@gmail.com> Co-authored-by: Keno Fischer <keno@juliacomputing.com> Co-authored-by: Jameson Nash <vtjnash@gmail.com> | 29 December 2020, 07:45:49 UTC |
4c4aaff | Keno Fischer | 29 December 2020, 05:03:26 UTC | Change arguments to (internal) `show_tuple_as_call` to kwargs (#39020) Forward port from #37849 to reduce the number of unrelated changes in that PR. The argument list here was simply getting unwieldy. The new `hasfirst` argument isn't yet used in this PR, but will be used in #37849. | 29 December 2020, 05:03:26 UTC |
a2b6426 | Keno Fischer | 29 December 2020, 02:57:55 UTC | Fix insert_node_here! if called at the end of the IR (#39021) This method was throwing an error when called after the compaction had processes the last instruction in the IR (with reverse affinity). This currently can't, because we never use this method after compacting a terminal instruction (inlining uses it to split BBs), but can after the new code in #37849, so fix the bug now. Forward port from #37849. | 29 December 2020, 02:57:55 UTC |
770e228 | Shuhei Kadowaki | 29 December 2020, 00:32:58 UTC | add news for `at-invokelatest` and `at-invoke` (#39000) | 29 December 2020, 00:32:58 UTC |
8e0183f | Mustafa M | 28 December 2020, 17:36:46 UTC | filesystem: code cleanup in request calls (#39012) The canonical way we allocate the libuv requests in the filesystem API is through malloc/free calls. Clean up the code to be more DRY. | 28 December 2020, 17:36:46 UTC |
187f96b | Max Horn | 28 December 2020, 14:35:01 UTC | Remove outdated references to Travis (#39011) | 28 December 2020, 14:35:01 UTC |
c3d7338 | Milan Bouchet-Valat | 28 December 2020, 14:26:48 UTC | Improve docs about let blocks (#39003) This section was probably written before the distinction between hard and soft scopes was added. | 28 December 2020, 14:26:48 UTC |
42bb1b2 | Mustafa M | 28 December 2020, 09:51:50 UTC | Revert "win: Set correct folder permissions after folder creation (#38942)" (#39014) This reverts commit 556cee7f0971407eaf547e291d66041e16293b6b. | 28 December 2020, 09:51:50 UTC |
471a55e | Jameson Nash | 27 December 2020, 18:02:35 UTC | Merge pull request #39009 from JuliaLang/sf/dlopen_libjulia Improve consistency of SONAME usage, fix debug installation | 27 December 2020, 18:02:35 UTC |
5d2fcdb | Aaron | 27 December 2020, 14:03:03 UTC | keys(::Generator) pass through (#34678) * keys(::Generator) pass-through Let keys pass through to allow lightweigth/lazy find* and arg* functions * Update test/functional.jl Co-authored-by: Matt Bauman <mbauman@gmail.com> Co-authored-by: Matt Bauman <mbauman@gmail.com> Co-authored-by: Simeon Schaub <simeondavidschaub99@gmail.com> | 27 December 2020, 14:03:03 UTC |
556cee7 | Mustafa M | 27 December 2020, 09:45:40 UTC | win: Set correct folder permissions after folder creation (#38942) libuv's mkdir does not implement the mode argument on Windows, as a result the folder mode is never passed on the Windows platform. Forcibly change the folder permissions to agree with the mode passed into `mkdir`. | 27 December 2020, 09:45:40 UTC |
d04793f | Simeon Schaub | 27 December 2020, 09:42:28 UTC | add missing newline to UnionAll Vararg warning (#39010) | 27 December 2020, 09:42:28 UTC |
176a6e6 | Elliot Saba | 27 December 2020, 00:43:44 UTC | Fix lack of `JL_MAJOR_SHLIB_EXT` on Windows | 27 December 2020, 00:43:44 UTC |
6c047cc | Elliot Saba | 26 December 2020, 23:58:57 UTC | [loader] Must invalidate `.o` files when `VERSION` changes (#38800) This doesn't happen that often, (only when `VERSION` changes) but we need to be sensitive to the path to `libjuliainternal.so` changing. | 26 December 2020, 23:58:57 UTC |
0af737c | Elliot Saba | 26 December 2020, 23:49:17 UTC | Use `strcmp()` to compare `f_lib` and `JL_LIBJULIA_DL_LIBNAME` (#39005) This future-proofs us from someone trying lookup a symbol from `libjulia.dll`, even though that should be quite rare, it may happen. | 26 December 2020, 23:49:17 UTC |
7aa128e | Elliot Saba | 26 December 2020, 23:35:07 UTC | Improve consistency of SONAME usage, fix debug installation SONAME and RPATH inconsistency resulted in `julia`/`julia-debug` occasionally being unable to find `libjulia.so`, due to `libjulia-internal` not having the right RPATH set (it was set to `$ORIGIN:$ORIGIN/julia`, when it now needs to be set to `$ORIGIN:$ORIGIN/..`). Use `patchelf` to address this during `make install`. We also want to use the SONAME of e.g. `libjulia` as much as possible, so as to avoid searching the filesystem for libraries that are already opened. Also fix a few bugs in building `julia-debug`. | 26 December 2020, 23:47:40 UTC |
d0478ca | Tim Holy | 26 December 2020, 20:33:01 UTC | Update the AST docs on `nothing` `:method` expressions (#38496) Fixes #38451 | 26 December 2020, 20:33:01 UTC |
8af7a04 | Jameson Nash | 22 December 2020, 04:04:01 UTC | build: pack checksums into fewer files by target (#38963) Hopefully this should preserve the ability to generate these in parallel (and keep a clean tree) and prevent conflicts when updating different deps targets, while consolidating any related files | 26 December 2020, 19:21:49 UTC |
70e00a6 | Steven G. Johnson | 26 December 2020, 18:30:35 UTC | move NEWS from #38675 to the correct section [ci skip] | 26 December 2020, 18:30:35 UTC |
729114f | Steven G. Johnson | 26 December 2020, 18:29:43 UTC | move `escape_string` NEWS (from #38597) to the correct section [ci skip] | 26 December 2020, 18:29:43 UTC |
dedf290 | Volker-Weissmann | 25 December 2020, 20:12:11 UTC | makefile: better debug comment in output (#39002) | 25 December 2020, 20:12:11 UTC |
a094c2f | Curtis Vogt | 25 December 2020, 17:23:25 UTC | Add reference to `indent` function (#38940) * Add reference to indent function * Update base/strings/io.jl Co-authored-by: Steven G. Johnson <stevenj@alum.mit.edu> Co-authored-by: Keno Fischer <keno@juliacomputing.com> Co-authored-by: Steven G. Johnson <stevenj@alum.mit.edu> | 25 December 2020, 17:23:25 UTC |
1e2e482 | Keno Fischer | 25 December 2020, 09:27:37 UTC | Remote restriction on documenting empty baremodules (#38998) Replaces and closes #13109. The underlying bug here appears to have been fixed and the restriction here was never triggered since lowering now always inserts LineNumberNodes, but we might as well remove the restriction since the bug is fixed. | 25 December 2020, 09:27:37 UTC |
560676d | Shuhei Kadowaki | 25 December 2020, 08:52:19 UTC | define at-invoke macro: (#38438) - provides easier syntax to call `Core.invoke`, e.g. `@invoke f(a::Integer)` will be expanded into `invoke(f, Tuple{Integer}, a)` - when type annotation is omitted, the argument type is specified as `Any` Built on top of #37971 | 25 December 2020, 08:52:19 UTC |
29d5d60 | Valentin Churavy | 25 December 2020, 07:35:59 UTC | Fix trampoline on PPC (#38980) * [Make] Normalize ppc64le to powerpc64le * [CLI] Fix trampoline on PowerPC 1. Add global entry 2. Don't store `r2` on parent frame 3. Conservative load from toc * remove jl_compile_extern_c from exported funcs | 25 December 2020, 07:35:59 UTC |
9fa97f9 | Volker-Weissmann | 25 December 2020, 07:35:05 UTC | fixed a bug in the build system (#38997) | 25 December 2020, 07:35:05 UTC |
ec386bd | Valentin Churavy | 25 December 2020, 02:04:19 UTC | Construct constant LLVMPtr correctly (#38958) Co-authored-by: Julian Samaroo <jpsamaroo@jpsamaroo.me> Co-authored-by: Julian Samaroo <jpsamaroo@jpsamaroo.me> | 25 December 2020, 02:04:19 UTC |
4d7289b | Katharine Hyatt | 25 December 2020, 02:01:44 UTC | Missing tests for accumulate (#38943) | 25 December 2020, 02:01:44 UTC |
570384e | Steven G. Johnson | 25 December 2020, 01:59:05 UTC | extend iseven/isodd to Number (#38976) * extend iseven/isodd to Number * add PR num to NEWS | 25 December 2020, 01:59:05 UTC |
60b68ce | Ronan Arraes Jardim Chagas | 25 December 2020, 01:50:31 UTC | Arg. to avoid escaping backslahes (escape_string) (#38597) * Arg. to avoid escaping backslahes (escape_string) This commit adds an argument to `escape_string` to assume that all backslashes (`\\`) are already escaped. Hence, `\\cdot` becomes `\\cdot` instead of `\\\\cdot`. The default value to this argument is `false`. Hence, this does not break any existing code since `escape_string` behaves exactly the same as before if no additional option is passed. Closes #34042 * Replace `bsescaped` arg. with `keep` keyword * Update NEWS.md * Fix doc tests * Add compat annotation Co-authored-by: Ronan Arraes Jardim Chagas <ronan.arraes@MacBook-Pro-de-Ronan.local> | 25 December 2020, 01:50:31 UTC |
6369f02 | Sebastian Stock | 25 December 2020, 01:49:35 UTC | Correct handling of non-Booleans in eval_test (#38695) | 25 December 2020, 01:49:35 UTC |
3aef83b | Simeon Schaub | 25 December 2020, 01:42:14 UTC | use parent type in similar(::ReinterpretedArray) (#38965) We already do this for `SubArray`, `PermutedDimsArray`, and `ReshapedArray`, so I think this makes sense here. This also adds tests for `similar(::ReshapedArray)`, since I didn't notice that was actually defined before. ref https://github.com/JuliaGPU/CUDA.jl/issues/602 | 25 December 2020, 01:42:14 UTC |
d70161e | Mustafa M | 25 December 2020, 01:37:42 UTC | Document __module and __source (#38987) Co-authored-by: June-6th <2660994113@qq.com> | 25 December 2020, 01:37:42 UTC |
390e40a | Katharine Hyatt | 25 December 2020, 01:36:33 UTC | Tests for push and pushfirst for abstractarray (#38991) | 25 December 2020, 01:36:33 UTC |
59c51f5 | Rik Huijzer | 25 December 2020, 01:35:47 UTC | Use occursin instead of findfirst(...) === nothing (#38990) | 25 December 2020, 01:35:47 UTC |
414410b | Rik Huijzer | 25 December 2020, 01:33:56 UTC | Document the opposites of any and all (#38985) | 25 December 2020, 01:33:56 UTC |
c377eed | Keno Fischer | 24 December 2020, 07:20:25 UTC | Refactor JL_GC_PUSHFRAME to internalize layout assumptions (#38982) Right now callers of JL_GC_PUSHFRAME need to be aware that the roots start three pointers after the end of the struct. That seems like an implementation detail of the macro, so refactor the macro to also take the assignment location for the locals instead. In the process fix an error where the gc-analyzer version of this code assumed a different layout. | 24 December 2020, 07:20:25 UTC |
d6a655b | Jameson Nash | 24 December 2020, 05:29:37 UTC | codegen: use new freeze intrinsics to avoid poison UB (#38977) LLVM would likely be required to insert this anyways to make hoisting legal (converting branches to and/or), so might as well add it ourselves, so that users can legally write that code directly. | 24 December 2020, 05:29:37 UTC |
ccaaf56 | Tim Holy | 24 December 2020, 05:12:02 UTC | Assert that _artifact_str returns a String (#38975) I have no idea whether this is important, but I noticed this while investigating triggers for inference in a package. Feel free to close if you think this is irrelevant. Incidentally, `_artifact_str` is a ~60ms precompile (that's just the inference, not the codegen). I see that its precompile was commented out in 924484f2f00; just mentioning it in case there's an opportunity to squeeze out a bit more latency-reduction. | 24 December 2020, 05:12:02 UTC |
48c4283 | Jameson Nash | 24 December 2020, 03:42:33 UTC | Merge pull request #38938 from imciner2/im/symfix Fix libjulia symbol trampolines, and export `jl_n_threads` | 24 December 2020, 03:42:33 UTC |