| 4e60912 | Katharine Hyatt | 03 March 2020, 07:30:28 UTC | More missing Hessenberg tests (#34966) | 03 March 2020, 07:30:28 UTC |
| f9efa88 | Katharine Hyatt | 03 March 2020, 07:24:51 UTC | Little test for ldiv error throw (#34964) | 03 March 2020, 07:24:51 UTC |
| 2f8cdd0 | Carsten Bauer | 02 March 2020, 20:40:40 UTC | Fix Base.show for SSHManager and LocalManager (#34955) | 02 March 2020, 20:40:40 UTC |
| 7588efb | willywalters5 | 02 March 2020, 20:36:45 UTC | help `;` now has output (#34944) | 02 March 2020, 20:36:45 UTC |
| 521813b | Kenta Sato | 02 March 2020, 20:35:50 UTC | test sprintf %p using real pointers (#34943) | 02 March 2020, 20:35:50 UTC |
| 51ced65 | oheil | 02 March 2020, 20:33:47 UTC | make Distributed independent of indices starting at 1 (#34886) | 02 March 2020, 20:33:47 UTC |
| 5b241c2 | Viral B. Shah | 02 March 2020, 14:46:33 UTC | arm -> armv7l in refresh-bb-tarballs.sh (#34935) Bump SuiteSparse from BB Add the armv7l binaries for openblas and remove checksums for older version | 02 March 2020, 14:46:33 UTC |
| 87e51d2 | Carsten Bauer | 02 March 2020, 14:37:46 UTC | Improve docstrings of IEEE floating point types (#34925) | 02 March 2020, 14:37:46 UTC |
| 9466232 | Rafael Fourquet | 02 March 2020, 12:20:30 UTC | add news/docs for #34524: 2-arg splice! on arbitrary index iterables (#34897) | 02 March 2020, 12:20:30 UTC |
| f709331 | Carsten Bauer | 02 March 2020, 05:51:43 UTC | Fix inverse of SVD of complex matrix (#34872) Closes #34866 | 02 March 2020, 05:51:43 UTC |
| 923d84b | Keno Fischer | 02 March 2020, 04:48:35 UTC | Revert "add readline(::AbstractCmd) (#34927)" (#34958) This reverts commit 570f6d9b2d3e72c4ce96d28b8f18f4c912b32d46. | 02 March 2020, 04:48:35 UTC |
| a523fcf | Keno Fischer | 01 March 2020, 22:55:15 UTC | Bump OpenBLAS (#34923) This has an experimental patch to hopefully help our Windows CI hangs. If this works, I'll need to include the openblas patch in the julia repo as well, but for now, let's just test if this helps CI. | 01 March 2020, 22:55:15 UTC |
| c2ce8f6 | Kristoffer Carlsson | 01 March 2020, 21:13:29 UTC | bump Pkg version (#34913) | 01 March 2020, 21:13:29 UTC |
| cdf580a | Kenta Sato | 01 March 2020, 16:49:01 UTC | fix unsigned(::Ptr) and signed(::Ptr) (#34941) | 01 March 2020, 16:49:01 UTC |
| 570f6d9 | Kenta Sato | 01 March 2020, 16:47:45 UTC | add readline(::AbstractCmd) (#34927) | 01 March 2020, 16:47:45 UTC |
| ea067fb | Johannes Ahnlide | 01 March 2020, 09:36:35 UTC | Fixed bug in reduced_index error message (#34937) | 01 March 2020, 09:36:35 UTC |
| ab28c9b | Katharine Hyatt | 01 March 2020, 09:33:17 UTC | Two missing tests for writedlm (#32984) | 01 March 2020, 09:33:17 UTC |
| 966a9b9 | Katharine Hyatt | 29 February 2020, 21:17:29 UTC | Some missing tests for Cholesky (#34929) * Test printing of CholeskyPivoted to REPL * Tests for type conversions * Fix factorization compare | 29 February 2020, 21:17:29 UTC |
| e39e945 | Shriprajwal K | 29 February 2020, 18:45:59 UTC | Remove unused single-argument _chol!(A::StridedMatrix) in cholesky.jl (#34850) | 29 February 2020, 18:45:59 UTC |
| 1bf5071 | Katharine Hyatt | 29 February 2020, 17:41:30 UTC | Merge pull request #34932 from crstnbr/cb/testsymmetric Add a missing test for Hermitian | 29 February 2020, 17:41:30 UTC |
| c4290fe | Kenta Sato | 29 February 2020, 17:36:35 UTC | define generic get! for AbstractDict (#34621) | 29 February 2020, 17:36:35 UTC |
| 4d8f889 | Justin Geeslin | 29 February 2020, 15:48:36 UTC | Documentation, dependency install command (macOS, Homebrew) (#32113) * Docs, Including command for installing gFortran, a dependency, via Homebrew * Line breaks and Formatting * Update doc/build/macos.md Co-Authored-By: Fredrik Ekre <ekrefredrik@gmail.com> * Update macos.md Co-authored-by: Viral B. Shah <viral@juliacomputing.com> Co-authored-by: Fredrik Ekre <ekrefredrik@gmail.com> | 29 February 2020, 15:48:36 UTC |
| b588181 | Katharine Hyatt | 29 February 2020, 15:06:46 UTC | Merge pull request #34930 from JuliaLang/ksh/sincods Add test for sincosd of missing | 29 February 2020, 15:06:46 UTC |
| 81bbfea | Katharine Hyatt | 29 February 2020, 14:27:30 UTC | Bunch of missing tests for Hessenberg (#34907) | 29 February 2020, 14:27:30 UTC |
| b1f3439 | Katharine Hyatt | 29 February 2020, 14:18:26 UTC | Some missing SVD tests (#34928) * Test for getting prop names of GeneralizedSVD * Test for REPL printing of GeneralizedSVD | 29 February 2020, 14:18:26 UTC |
| e9b7060 | Katharine Hyatt | 29 February 2020, 14:01:11 UTC | Missing diag tests for block SymTriDiag (#34926) | 29 February 2020, 14:01:11 UTC |
| 7533d04 | Katharine Hyatt | 29 February 2020, 12:35:46 UTC | LU to Factorization tests (#34905) | 29 February 2020, 12:35:46 UTC |
| e453cdb | Katharine Hyatt | 29 February 2020, 06:33:34 UTC | Add test for sincosd of missing | 29 February 2020, 06:33:34 UTC |
| 17b2776 | Carsten Bauer | 29 February 2020, 00:00:11 UTC | added a test_throws for Hermitian | 29 February 2020, 00:00:11 UTC |
| 6dd9f41 | jmert | 28 February 2020, 22:34:56 UTC | Fix assignment to 0-dim array indexed by CartesianIndices{0} (#34893) * Fix assignment to 0-dim array indexed by CartesianIndices{0} * Fix assigning element into ≥1-dim arrays from 0-dim array | 28 February 2020, 22:34:56 UTC |
| 3f05b0d | Keno Fischer | 28 February 2020, 21:03:34 UTC | Fix nothrow behavior for intrinsics (#34539) This fixes the case in #34482 and adds some more robustness for similar cases, though I wouldn't be surprised if there were more dragons hiding here. Intrinsics are a bit special and from the start, we sort of expected them to only ever be called correctly under pain of segfaults or other undefined behavior. We've been gradually making these more robust, but fundamentally, they were never intended to be used by users directly, only through the type-validating wrappers in Base. What has changed to cause the recent influx of issues in this area is that people now like to do compiler transforms that happily recurse through these wrappers and perform transforms that are not always legal on Intrinsics. This should help catch a number of them, but this interface is still not very thoroughly validated, and I would be surprised to see crashes or other errors stemming from incorrect usage here. | 28 February 2020, 21:03:34 UTC |
| 012b270 | Katharine Hyatt | 28 February 2020, 07:57:24 UTC | More tests for Bidiagonal (#34901) * Tranpose for non-numeric eltype bidiag tests * More tests for bidiag mul | 28 February 2020, 07:57:24 UTC |
| 1c24d30 | Aaron Christianson | 28 February 2020, 04:31:19 UTC | A more julian syntax for ccall: the @ ccall macro (#32748) Here we implement a syntax for ccall with Julia-like type annotations on the arguments. Compared to ccall: * The new syntax is more familiar to Julia programmers * The new syntax gives a notation for calling C varargs functions * Support for specifying calling convention is not yet implemented as that will require another syntax discussion * The semantics for interpolating "function like things" is much simplified (only function pointers are allowed) | 28 February 2020, 04:31:19 UTC |
| 6abc852 | Keno Fischer | 28 February 2020, 03:34:01 UTC | Fix regression in inlining of invoke (#34906) When we added the check to prevent inlining through ambiguous methods, we failed exclude this check for calls to `invoke` (which skip ambiguous methods, since the method is explicitly selected). Fixes #34900. | 28 February 2020, 03:34:01 UTC |
| aad9664 | Katharine Hyatt | 27 February 2020, 22:17:25 UTC | Merge pull request #34899 from aviatesk/avi/fixdocsyntax fix doc syntax of `Channel` | 27 February 2020, 22:17:25 UTC |
| e0965dd | Shuhei Kadowaki | 27 February 2020, 16:38:46 UTC | fix doc syntax of `Channel` | 27 February 2020, 16:38:46 UTC |
| f0636b2 | Katharine Hyatt | 27 February 2020, 13:55:27 UTC | More tests for Triangular, div and mult (#31831) * More tests for Triangular, div and mult * Long form function defs * avoid useless unwrapping-wrapping in triangular multiplication Co-authored-by: Daniel Karrasch <Daniel.Karrasch@gmx.de> | 27 February 2020, 13:55:27 UTC |
| 57a9821 | Rafael Fourquet | 27 February 2020, 10:26:28 UTC | fix docstring of `replace` | 27 February 2020, 10:26:28 UTC |
| a91edd0 | Pablo San-Jose | 27 February 2020, 09:44:38 UTC | no-op for copyto!(dst, emptysrc) (#34891) | 27 February 2020, 09:44:38 UTC |
| 448cbfc | Pablo San-Jose | 27 February 2020, 09:39:30 UTC | no-op copyto!(dense, emptysparse) with any eltype (#34892) | 27 February 2020, 09:39:30 UTC |
| c2cd601 | Keno Fischer | 26 February 2020, 20:41:26 UTC | Allow 2-arg splice! on arbitrary index iterables (#34524) We have deleteat!(vec::Vector, itr), but it returns the modified vector. If you instead want to obtain the delted elements, there is splice!(::Vector, ::UnitRange) as a special case of the three-argument version that inserts additional elements where the old ones were deleted. However, these is no two argument splice! for arbitrary iterables, so probably the best way to write that currently is: ``` inds = collect(itr) vals = A[inds] deleteat!(A, inds) vals ``` which is both less efficient and more verbose than ideal. I'm wondering if perhaps deleteat! should have been returning the deleted elements, but that's not a change we can make in 1.x. Instead, I'm proposing here to extend the 2 argument (but not the 3 argument for obvious reasons) variant of splice! to arbitrary iterables. | 26 February 2020, 20:41:26 UTC |
| c1059e7 | Nathan Daly | 26 February 2020, 20:31:59 UTC | Change TypeVars in UnionAlls to print as var"" syntax (#34888) This changes unnamed type variables to print as valid syntax, via the `var""` printing macro. For example, before this commit, `Vector{<:Bool}` printed as a non-syntactic string: ```julia julia> Vector{<:Bool} Array{#s16,1} where #s16<:Bool ``` And now it prints as a valid, syntactic expression: ```julia julia> Vector{<:Bool} Array{var"#s46",1} where var"#s46"<:Bool ``` | 26 February 2020, 20:31:59 UTC |
| fde1ec8 | Takafumi Arakaki | 26 February 2020, 20:08:57 UTC | Implement `accumulate` and friends for Tuple (#34654) | 26 February 2020, 20:08:57 UTC |
| b92d678 | Chris Foster | 26 February 2020, 20:05:05 UTC | Add file name to all top level line number nodes (#34881) This allows julia code to more easily consume the output of jl-parse-all (see #34715) and is consistent with the way line number nodes have file name information in all non-toplevel constructs. | 26 February 2020, 20:05:05 UTC |
| 9f680d3 | Shriprajwal K | 26 February 2020, 17:01:23 UTC | Added test for the seekend function in SecretBuffer (#34731) | 26 February 2020, 17:01:23 UTC |
| 808ec9c | Shriprajwal K | 26 February 2020, 17:00:37 UTC | Added tests for hashing of SecretBuffer (#34736) | 26 February 2020, 17:00:37 UTC |
| 434c6bb | Andreas Noack | 26 February 2020, 09:01:16 UTC | Loosen test of Cholesky factorization of diagonal matrices to approximate (#34859) equality since BLAS and Julia might use square roots that round differently. This has occasionally caused CI failures. | 26 February 2020, 09:01:16 UTC |
| 0fe45ef | Keno Fischer | 25 February 2020, 23:15:03 UTC | Revert "Added test for dumptype with a Concrete and Abstract type. (#34764)" (#34878) This reverts commit 4ab8a9d27fb608f287f63cd6ebe478aef15500c5. | 25 February 2020, 23:15:03 UTC |
| c934b94 | Jeff Bezanson | 25 February 2020, 17:15:16 UTC | timing.c needs to include `julia_internal.h` for `cycleclock` (#34861) The function was moved by 65b8e7ed72541d1dd5bc1f4dcabc5245e1dba718 | 25 February 2020, 17:15:16 UTC |
| d559015 | Shriprajwal K | 25 February 2020, 15:42:32 UTC | Added SecretBuffer test for the skip function (#34740) | 25 February 2020, 15:42:32 UTC |
| 862e0fe | Jameson Nash | 25 February 2020, 15:41:10 UTC | gf: optimize jl_matching_methods for dispatch tuples (#34745) When the input type is a dispatch tuple, we know that a subtype lookup is sufficient, so we can do a slightly more direct lookup and avoid considering `Union{}`-declared methods as possible results. This also fixes an issue with the special `jl_matching_methods` query being done by `jl_insert_backedges`: we would set the `done` flag when a method "fully covered" the result, even if that method was deleted. Rather than continue to abuse the meaning of `world == 0` (which also has performance implications), we add a specific flag for this case. | 25 February 2020, 15:41:10 UTC |
| 1bd0593 | Rafael Fourquet | 25 February 2020, 14:39:23 UTC | faster BigInt hashing (#33790) * faster BigInt hashing | 25 February 2020, 14:39:23 UTC |
| ef0e0f2 | Jeff Bezanson | 24 February 2020, 22:36:53 UTC | remove special cases for `Main` in system image saving (#34828) `usings` and all bindings will now be preserved make Core a top-level module (its parent is itself) instead of setting its parent to Main. | 24 February 2020, 22:36:53 UTC |
| 7bc7874 | Jameson Nash | 11 February 2020, 21:04:48 UTC | gf: optimize jl_matching_methods for dispatch tuples When the input type is a dispatch tuple, we know that a subtype lookup is sufficient, so we can do a slightly more direct lookup and avoid considering `Union{}`-declared methods as possible results. This also fixes an issue with the special `jl_matching_methods` query being done by `jl_insert_backedges`: we would set the `done` flag when a method "fully covered" the result, even if that method was deleted. Rather than continue to abuse the meaning of `world == 0` (which also has performance implications), we add a specific flag for this case. | 24 February 2020, 20:02:32 UTC |
| 9af0715 | Ravishankar Joshi | 24 February 2020, 19:32:29 UTC | Added test for modifying a tuple element using setindex (#34857) | 24 February 2020, 19:32:29 UTC |
| f0f563e | bramtayl | 24 February 2020, 19:20:06 UTC | improve fastmath test coverage (#34702) | 24 February 2020, 19:20:06 UTC |
| 6f3c5ce | kczimm | 24 February 2020, 18:44:10 UTC | Add show test for MIME type tab-separated-values (#34766) | 24 February 2020, 18:44:10 UTC |
| d4ac457 | Logan Kilpatrick | 24 February 2020, 18:38:45 UTC | Add README link to learning resources (#34751) | 24 February 2020, 18:38:45 UTC |
| 909422d | Shriprajwal K | 24 February 2020, 18:37:05 UTC | Added test for bytes available in a SecretBuffer (#34762) | 24 February 2020, 18:37:05 UTC |
| e29a1aa | Ryan Birmingham | 24 February 2020, 18:35:27 UTC | add filesystem tests to increase coverage (#34831) | 24 February 2020, 18:35:27 UTC |
| 4ab8a9d | kczimm | 24 February 2020, 18:34:13 UTC | Added test for dumptype with a Concrete and Abstract type. (#34764) | 24 February 2020, 18:34:13 UTC |
| 5844117 | Chris Foster | 24 February 2020, 18:14:04 UTC | Remove spurious warnings with g++-7.4.0 (#34858) * Complaints about "potentially uninitialized" pointers * Acknowledge LLVM ORCJIT v1 deprecations | 24 February 2020, 18:14:04 UTC |
| 835e148 | Daniel Karrasch | 24 February 2020, 11:26:07 UTC | Add multiplication/division with/by Periods (#34786) | 24 February 2020, 11:26:07 UTC |
| 39b431d | Pablo San-Jose | 24 February 2020, 08:47:47 UTC | Specialize copyto!(dense, sparse) (#33589) | 24 February 2020, 08:47:47 UTC |
| 5286b40 | cormullion | 23 February 2020, 16:29:10 UTC | div.jl - divrem - unneeded "the" removed (#34848) | 23 February 2020, 16:29:10 UTC |
| 3fdfb67 | Pablo San-Jose | 23 February 2020, 13:49:53 UTC | Consistently use linear indexing in copyto!(dest, src) (#34049) | 23 February 2020, 13:49:53 UTC |
| 8e6abe2 | Tamas K. Papp | 23 February 2020, 11:08:54 UTC | Clarify that eachrow and eachcol return vectors, add examples. (#34845) | 23 February 2020, 11:08:54 UTC |
| ba96e8b | Valentin Churavy | 22 February 2020, 11:42:41 UTC | Merge pull request #34760 from thomasfaingnaert/tf/as-ptr Add AddrSpacePtr type | 22 February 2020, 11:42:41 UTC |
| 14b4528 | Katharine Hyatt | 22 February 2020, 10:20:14 UTC | Fix maybe-bug in sparsevec, more tests (#34827) | 22 February 2020, 10:20:14 UTC |
| 8cb8f0c | Thomas Faingnaert | 14 February 2020, 14:01:04 UTC | Add AddrSpacePtr type | 21 February 2020, 23:04:14 UTC |
| 3f0b8c9 | Tim Holy | 21 February 2020, 22:05:04 UTC | Merge pull request #34770 from JuliaLang/teh/oa92 Convert range type in `reduced_index` | 21 February 2020, 22:05:04 UTC |
| dc46ddd | Jameson Nash | 21 February 2020, 19:20:07 UTC | Merge pull request #32599 from JuliaLang/jn/32511-optim threads: further optimize scheduler based on #32551 | 21 February 2020, 19:20:07 UTC |
| f5dbc47 | Steven G. Johnson | 21 February 2020, 19:09:58 UTC | NamedTuple macro for easier type construction (#34548) | 21 February 2020, 19:09:58 UTC |
| c17be46 | Peng Guanwen | 21 February 2020, 16:19:31 UTC | Add documentation for takewhile and dropwhile (#34837) #33437 has added the two functions, but it didn't appear in the documentation. | 21 February 2020, 16:19:31 UTC |
| 3631d91 | Jeff Bezanson | 21 February 2020, 16:05:38 UTC | avoid leaving closure type from sysimg.jl in Main (#34825) | 21 February 2020, 16:05:38 UTC |
| 5cd7482 | Jeff Bezanson | 21 February 2020, 16:05:00 UTC | fix dict `x == x` to return missing if `x` contains it (#34809) closes #34744 use `isequal` to compare keys in `ImmutableDict` | 21 February 2020, 16:05:00 UTC |
| a3f7fa8 | Jeff Bezanson | 21 February 2020, 16:02:10 UTC | fix #34826, unwrap VecElement when needed in `julia_const_to_llvm` (#34829) | 21 February 2020, 16:02:10 UTC |
| 8e3aa66 | Sebastian Micluța-Câmpeanu | 21 February 2020, 16:00:05 UTC | Fix #34316 (#34817) Fix hypot with both arguments 0 Remove hypot from Furlongs in order to test the defined fallback | 21 February 2020, 16:00:05 UTC |
| eac7411 | Jeffrey Lin | 21 February 2020, 15:58:46 UTC | Add bitrotate function (issue #11592) (#33937) | 21 February 2020, 15:58:46 UTC |
| d8b2209 | László Papp | 21 February 2020, 08:38:02 UTC | Add spmv! to BLAS in stdlib/LinearAlgebra (#34320) | 21 February 2020, 08:38:02 UTC |
| 5ecd1cf | Kristoffer Carlsson | 21 February 2020, 08:02:45 UTC | remove trim option to fkeep! and related functions (#32972) * remove trim option to fkeep! and related functions * add NEWS entry | 21 February 2020, 08:02:45 UTC |
| 8009429 | Morten Piibeleht | 20 February 2020, 22:14:03 UTC | Don't use startup.jl when building docs (#34795) | 20 February 2020, 22:14:03 UTC |
| 0d39976 | Daniel Karrasch | 20 February 2020, 21:26:24 UTC | fix build issues introduced in #28872 (#34819) | 20 February 2020, 21:26:24 UTC |
| 41f07ac | Kristoffer Carlsson | 20 February 2020, 20:55:25 UTC | remove error when calling fieldtype on a Module (#34804) | 20 February 2020, 20:55:25 UTC |
| 7e2363b | Takafumi Arakaki | 20 February 2020, 19:43:04 UTC | Add \bbsemi as a binary operator (#34722) | 20 February 2020, 19:43:03 UTC |
| 6476458 | Mohamed Tarek | 20 February 2020, 19:36:32 UTC | zero and one for irrational numbers (#34773) | 20 February 2020, 19:36:32 UTC |
| fb301f4 | Jeff Bezanson | 20 February 2020, 19:33:08 UTC | add `bitreverse` function (#34791) | 20 February 2020, 19:33:08 UTC |
| 6a9a926 | László Papp | 20 February 2020, 07:47:25 UTC | Blas hpmv minor updates (#34439) | 20 February 2020, 07:47:25 UTC |
| e803d78 | Dilum Aluthge | 19 February 2020, 22:57:10 UTC | Documentation: Mention `Vararg{Any, N}` in the "Performance Tips" section of the manual (#34796) | 19 February 2020, 22:57:10 UTC |
| 65b8e7e | Jameson Nash | 10 July 2019, 22:47:06 UTC | threads: further optimize scheduler wake-ups And add some debugging aids, controlled by changing the `JULIA_DEBUG_SLEEPWAKE()` macro in `julia_threads.h` Eliminates the `sleep_check_state` global, equal to `any(ptls->sleep_check_state == true)` | 19 February 2020, 19:10:18 UTC |
| c23554a | Jameson Nash | 20 January 2020, 22:31:46 UTC | partr: fix multiqueue resorting stability | 19 February 2020, 19:10:18 UTC |
| 2a35739 | Jameson Nash | 13 September 2019, 17:53:22 UTC | sigusr1: restore ability to dump program state on linux signal | 19 February 2020, 19:10:18 UTC |
| 3bcedb2 | Curtis Vogt | 19 February 2020, 19:09:51 UTC | Remove reference to build_sysimg.jl (#34765) | 19 February 2020, 19:09:51 UTC |
| 4bba0aa | Jameson Nash | 19 February 2020, 19:09:25 UTC | various improvements to subtyping (#34743) * subtype: optimize even if envsz > 0 * typeintersect: fix bug with non-type and Any * intersect: skip intersection test for more concrete cases * obvious_subtype: most supertypes of Type are obvious | 19 February 2020, 19:09:25 UTC |
| f5afdf9 | Jameson Nash | 19 February 2020, 18:51:30 UTC | Profile: fix indent computation BoundsError (#34806) fixes #34723 | 19 February 2020, 18:51:30 UTC |
| 8720530 | Jameson Nash | 19 February 2020, 18:50:55 UTC | codegen: fix segfaults with VecElement with union element (#34805) We were being too aggressive at unwrapping these: since they are already unwrapped once, they need to be treated almost as a normal struct. fixes #29864 | 19 February 2020, 18:50:55 UTC |
| d634f73 | Jameson Nash | 19 February 2020, 18:43:38 UTC | codegen: fix bad declaration of memcmp stdlib function signature (#34792) We were lying to LLVM, and it now catches that in the verifier. Stop lying to LLVM so that it will again be happy to generate good code for us (e.g. optimizing this to bcmp). | 19 February 2020, 18:43:38 UTC |
| f36edc2 | Keno Fischer | 19 February 2020, 18:20:48 UTC | partr: Fix deadlock (#34807) When there is no work to do, the first thread to be idle will attempt to run the event loop once, waiting for any notifications (which will usually create new work). However, there is an interesting corner case where a notification arrives, but no work was scheduled. That doesn't usually happen, but there are a few situations where it does: 1) Somebody used a libuv primitive outside of julia, so the callback doesn't schedule any julia work. 2) Another thread forbily interrupted the waiting thread because it wants to take over the event loop for various reasons 3) On Windows, we occasionally get spurious wake ups of the event loop. The existing code in partr assumed that we were in situation 2, i.e. that there was another thread waiting to take over the event loop, so it released the event loop and simply put the current thread to sleep in the expectation that another thread will pick it up. However, if we instead are in one of the other two conditions, there may not be another thread there to pick up the event loop. Thus, with no thread owning the event loop, julia will stop responding to events and effectively deadlock. Since both 1 and 3 are rare, and we don't actually enter the event loop until there was no work for 4 milliseconds (which is fairly rare), this condition rarely happens, but is occasionally observable on Windows, where it caused #34769. To test that this fix works, we manually create situation 1 in the test by creating an idle callback, which will prevent the event loop from blocking, but only schedules julia work after it's been called 100 times. This reproduces the observed failure from the issue and is fixed by this PR. Fixes #34769 Co-authored-by: Jeff Bezanson <jeff@juliacomputing.com> Co-authored-by: Jameson Nash <jameson@juliacomputing.com> | 19 February 2020, 18:20:48 UTC |
| 4188b2c | Zepp | 19 February 2020, 14:24:18 UTC | Add some (matrix) functions for UniformScaling (#28872) Co-authored-by: Daniel Karrasch <Daniel.Karrasch@gmx.de> | 19 February 2020, 14:24:18 UTC |
| 6ade90a | Andreas Noack | 19 February 2020, 11:33:46 UTC | Fix geqp3! documentation (#34784) | 19 February 2020, 11:33:46 UTC |
