| 236df47 | Valentin Churavy | 14 May 2019, 00:03:42 UTC | Merge pull request #31957 from wsmoses/master Fix compilation with external LLVM | 14 May 2019, 00:03:42 UTC |
| 178d703 | Lyndon White | 13 May 2019, 20:54:59 UTC | Split rule in Generated function docs about calling functions from argument types (#32016) | 13 May 2019, 20:54:58 UTC |
| 650a6a3 | Valentin Churavy | 13 May 2019, 16:38:35 UTC | Merge pull request #32014 from JuliaLang/vs/rmllvm39 Remove old llvm 3.9 checksums | 13 May 2019, 16:38:35 UTC |
| 3efdda6 | Mosè Giordano | 13 May 2019, 09:08:23 UTC | fast implementation of digits with base a power of 2 (#31722) | 13 May 2019, 09:08:23 UTC |
| 5dc47ae | Viral B. Shah | 13 May 2019, 04:26:20 UTC | Remove vagrant files in contrib that were left around. | 13 May 2019, 04:26:20 UTC |
| 531c2d8 | Viral B. Shah | 13 May 2019, 04:24:49 UTC | Remove old llvm 3.9 checksums | 13 May 2019, 04:24:49 UTC |
| 9d6fffe | Mustafa M | 12 May 2019, 17:20:46 UTC | Improve some aspects of the NSIS installer (#31959) | 12 May 2019, 17:20:46 UTC |
| 7fdd96f | lbilli | 12 May 2019, 17:05:33 UTC | Fix function name in docstring (#32009) | 12 May 2019, 17:05:33 UTC |
| 8932ccd | Tamas K. Papp | 12 May 2019, 11:21:32 UTC | document Random samplers: continuation of #31787 (#31990) | 12 May 2019, 11:21:32 UTC |
| bfcf172 | Jeff Bezanson | 11 May 2019, 15:50:04 UTC | fix #31993, intersection of `Type{Vec}` and `Type{<:Vec{T}}` (#31997) | 11 May 2019, 15:50:04 UTC |
| 40296fb | Keno Fischer | 11 May 2019, 15:44:25 UTC | Don't attempt to const prop call cycles (#31996) Even if the result is unused. Fixes #31974. | 11 May 2019, 15:44:25 UTC |
| 7f2fa84 | Daniel Karrasch | 10 May 2019, 21:35:04 UTC | doc: clarify docstring of lmul!/rmul! (#31282) | 10 May 2019, 21:35:04 UTC |
| f0ffb29 | Jeff Bezanson | 10 May 2019, 20:06:05 UTC | synchronize FileWatching (#31981) | 10 May 2019, 20:06:05 UTC |
| 25fc3a9 | Jeff Bezanson | 10 May 2019, 20:05:01 UTC | fix synchronization to ensure a task can only run on one thread at a time (#31619) * fix synchronization to ensure a task can only run on one thread at a time * fix hang in I/O blocking with non-sticky tasks If a thread's scheduler is running on a certain task, only add that task to its sticky queue and not the shared queue, since only that thread can run it anyway. | 10 May 2019, 20:05:01 UTC |
| 779ac77 | Elliot Saba | 10 May 2019, 16:23:52 UTC | Fix for optind in getopt on musl libc. (#31946) Fix for optind in getopt on musl libc. | 10 May 2019, 16:23:52 UTC |
| 6f463da | Elliot Saba | 10 May 2019, 15:55:43 UTC | Add GCC 9 support to `normalize_triplet.py` (#31976) | 10 May 2019, 15:55:42 UTC |
| 5486cc2 | Rafael Fourquet | 10 May 2019, 14:17:56 UTC | fix updated show test in #31087 (#31991) | 10 May 2019, 14:17:56 UTC |
| 8d2727b | Keno Fischer | 10 May 2019, 13:41:59 UTC | Accept both Cvoid and Ptr{Void} return type in ccall(:memcpy) (#31464) POSIX specifies that memcpy returns its first argument, which is a bit of a useless feature (e.g. the llvm intrinsic just returns `void`. Nevertheless, since we're intercepting the ccall here, we should allow it. For convenience, still allow the Cvoid return though. Fixes #31073 | 10 May 2019, 13:41:59 UTC |
| 9e1b368 | Katharine Hyatt | 10 May 2019, 13:33:52 UTC | Some missing bidiag tests (#31888) | 10 May 2019, 13:33:52 UTC |
| 829a4a6 | Chris Foster | 09 May 2019, 22:10:18 UTC | Fix pop_exception lowering for try-break-finally (#31766) (#31931) In the case where a finally handler was active, pop_exception was incorrectly omitted when lowering `break` inside a catch block. This leads to stale exceptions on the stack. | 09 May 2019, 22:10:18 UTC |
| decb802 | Jameson Nash | 09 May 2019, 20:05:19 UTC | sysimg: reduce the number of special cases (#31948) Put the entire image(*) into our pre-constructed data image, instead of attempting to leave out a couple of items. This saves us a couple lines of code, and should makes it simpler to keep the list of builtin objects (gc.c "tags") up-to-date. (*) The remaining special objects are ptls->root_task, since I was unsure the best way to handle it as no other Task object is allowed to be referenced, and the small integer caches. | 09 May 2019, 20:05:19 UTC |
| e3afef0 | Jeff Bezanson | 09 May 2019, 19:00:25 UTC | fix #31965, bug in interpreter stacktraces and `jl_code_requires_compiler` (#31967) | 09 May 2019, 19:00:25 UTC |
| bf58070 | Elliot Saba | 09 May 2019, 18:42:19 UTC | Change BB tarball naming scheme to include release number (#31769) Change BB tarball naming scheme to include release number | 09 May 2019, 18:42:19 UTC |
| 14b74bc | Keno Fischer | 09 May 2019, 16:24:41 UTC | Fix world age issues in Serialization test if run in Main (#31973) Fixes #31966 | 09 May 2019, 16:24:41 UTC |
| e813f0d | Tamas K. Papp | 09 May 2019, 11:01:27 UTC | Expand documentation of custom random samplers. (#31787) | 09 May 2019, 11:01:27 UTC |
| 48634f9 | Jeff Bezanson | 08 May 2019, 23:19:20 UTC | always run on original process stack in ALWAYS_COPY_STACKS mode (#31697) this makes it possible to work around #31104 | 08 May 2019, 23:19:20 UTC |
| a526662 | Chris Foster | 08 May 2019, 19:58:04 UTC | Revert "Don't use jl_rethrow_other for LoadError/InitError" (#31963) This reverts commit d3dbe86f49da6779c2f3af8354c6a3933f48fcad. + fixup new tests in backtrace.jl | 08 May 2019, 19:58:04 UTC |
| 4c28b36 | Jeff Bezanson | 08 May 2019, 18:38:34 UTC | fix #31899, type intersection involving Int in upper bound (#31960) | 08 May 2019, 18:38:34 UTC |
| c9777b0 | Jameson Nash | 08 May 2019, 14:34:40 UTC | generate_precompile script: try to improve reliability (#30040) TTY objects are inherently unreliable input channels (documented, per posix design), so try to slow down the rate of input and speed up the rate of output by watching the output stream more closely, and detecting intermediate errors. And also just general cleanup some of our IO handling. | 08 May 2019, 14:34:40 UTC |
| 5afc335 | William S. Moses | 07 May 2019, 18:35:10 UTC | Fix compilation with external LLVM | 07 May 2019, 18:35:10 UTC |
| b126788 | Elliot Saba | 07 May 2019, 14:45:21 UTC | Merge pull request #31900 from musm/patch-16 Remove unused download of 7z-extra | 07 May 2019, 14:45:21 UTC |
| a6c7c1b | Jameson Nash | 06 May 2019, 22:53:38 UTC | Merge pull request #31877 from JuliaLang/jn/lazy-ftypes types: lazy initialize the field-types when first needed | 06 May 2019, 22:53:38 UTC |
| 40ce424 | Jameson Nash | 06 May 2019, 22:49:25 UTC | Merge pull request #31942 from JuliaLang/revert-31895-jn/append_any Re-merge commit for "optimize append_any more" (#31895) | 06 May 2019, 22:49:25 UTC |
| 9de07ce | Fredrik Ekre | 06 May 2019, 13:55:32 UTC | Fix optind in getopt on musl libc. On exit optind points to the last non-opt argument of argv, but in the case where there are only options optind does not go beyond argc, except on musl libc, where it becomes argc + 1. | 06 May 2019, 22:18:12 UTC |
| 11ce4d1 | Fredrik Ekre | 06 May 2019, 20:43:15 UTC | Fix failing doctests from PR #31246. (#31934) | 06 May 2019, 20:43:15 UTC |
| 4770b7e | Mustafa M | 06 May 2019, 20:14:20 UTC | Simplify Appveyor build script (#31917) | 06 May 2019, 20:14:20 UTC |
| 040a3e5 | Fredrik Ekre | 06 May 2019, 19:01:37 UTC | Unconditionally append lib/julia to rpath, fixes #31903. (#31933) | 06 May 2019, 19:01:37 UTC |
| aa7ae5a | Fredrik Ekre | 06 May 2019, 19:00:15 UTC | Make --code-coverage testing more robust for filepath. (#31939) | 06 May 2019, 19:00:15 UTC |
| 1b2d3ca | Jameson Nash | 01 May 2019, 16:57:30 UTC | apply: unify append_any implementations Rather than optimize two copies of the same function, we can just have one version of the function that handles everything. | 06 May 2019, 16:43:28 UTC |
| 4ae4348 | Jameson Nash | 01 May 2019, 16:43:57 UTC | inference: improve sizeof tfunc Primitive types also always have a sizeof. This is particularly relevant for `Ptr`, since `elsize` uses `sizeof(Ptr)` to compute the stride. | 06 May 2019, 16:43:25 UTC |
| f3312ce | Jameson Nash | 06 May 2019, 16:39:49 UTC | Revert "optimize append_any more" | 06 May 2019, 16:39:49 UTC |
| 9e9bcb8 | Jameson Nash | 06 May 2019, 16:37:21 UTC | optimize append_any more (#31895) * inference: improve sizeof tfunc Primitive types also always have a sizeof. This is particularly relevant for `Ptr`, since `elsize` uses `sizeof(Ptr)` to compute the stride. * apply: unify append_any implementations Rather than optimize two copies of the same function, we can just have one version of the function that handles everything. | 06 May 2019, 16:37:21 UTC |
| 7ef9e75 | Jameson Nash | 06 May 2019, 16:34:48 UTC | build: minor cleanup of ui/Makefile (#31919) Adds some prereqs, and makes more use of existing variables to get our desired results. Fixes #31876 Replaces #31904 Co-Authored-By: Jameson Nash <vtjnash@gmail.com> | 06 May 2019, 16:34:48 UTC |
| bdffb56 | Mustafa M | 06 May 2019, 13:58:49 UTC | Improve edit when editor set to Visual Studio code (#31864) By default VS Code adds `code.cmd` to path. Allow `code.cmd` within the vs code chain. | 06 May 2019, 13:58:49 UTC |
| f189ae4 | Elliot Saba | 06 May 2019, 04:23:38 UTC | Use https for downloads of 7z and NSIS and also add checksums. (#31902) Use https for downloads of 7z and NSIS and also add checksums. | 06 May 2019, 04:23:38 UTC |
| a0cee0b | Mustafa M | 06 May 2019, 01:35:12 UTC | Merge branch 'master' into patch-16 | 06 May 2019, 01:35:12 UTC |
| 3736c1a | Mus | 02 May 2019, 19:30:18 UTC | Improve downloads in win-extras Use hhtps for the downloads and also add checksums | 06 May 2019, 01:33:23 UTC |
| 9b5fea5 | Jan Weidner | 04 May 2019, 12:32:06 UTC | Docstrings for quot, isexpr and show_sexpr (#31246) | 04 May 2019, 12:32:06 UTC |
| c40d9b0 | Stefan Karpinski | 04 May 2019, 03:30:22 UTC | README: move "Binary Installation" up and tidy it (#31912) Since this is how the vast majority of people will want to install Julia, it seems sensible to have it near the top of the page. This section had some very weird formatting in terms of whitespace, so I tidied that up while I was at it. | 04 May 2019, 03:30:22 UTC |
| b16345d | Steven G. Johnson | 04 May 2019, 02:05:56 UTC | use Base._memcmp instead of ccall(:memcmp) (#31906) * use Base._memcmp instead of ccall(:memcmp) * Update base/strings/string.jl Co-Authored-By: stevengj <stevenj@mit.edu> | 04 May 2019, 02:05:56 UTC |
| a086ad8 | Timo Kluck | 04 May 2019, 00:04:52 UTC | Micro-optimization: remove unnecessary allocation for getindex(::Number, i::Integer...) (#31918) | 04 May 2019, 00:04:52 UTC |
| b96a591 | Valentin Churavy | 03 May 2019, 20:01:18 UTC | Merge pull request #31849 from JuliaLang/jn/llvm-up llvm upgrade: apply applicable old patches | 03 May 2019, 20:01:18 UTC |
| a7fdfdb | Jameson Nash | 26 April 2019, 17:30:43 UTC | llvm: apply applicable old patches for 7 and 8 also disabled a new X86 pass (x86-disable-avoid-SFB) to avoid running into an assertion error in that pass (see gist) | 03 May 2019, 13:45:28 UTC |
| bd95f03 | Jameson Nash | 28 April 2019, 01:49:17 UTC | codegen: fix tbaa info for select intrinsic, review other tbaa annotations | 03 May 2019, 13:45:28 UTC |
| ba3a1f8 | Katharine Hyatt | 03 May 2019, 13:31:07 UTC | Merge pull request #31887 from JuliaLang/ksh/dense Three missing dense tests | 03 May 2019, 13:31:07 UTC |
| 4111609 | Jameson Nash | 03 May 2019, 13:05:27 UTC | improve compilation of check_top_bit (#31883) addresses #31819 | 03 May 2019, 13:05:27 UTC |
| 2fb61c3 | Elliot Saba | 03 May 2019, 07:07:34 UTC | Revert "Upgrade and use NSIS for compression instead of extra 7z stag… (#31907) Revert "Upgrade and use NSIS for compression instead of extra 7z stag… | 03 May 2019, 07:07:34 UTC |
| 289d5b6 | Mus | 03 May 2019, 04:26:31 UTC | Revert "Upgrade and use NSIS for compression instead of extra 7z stage (#31879)" This reverts commit b118e68a549a2294f70889390e49b0a8ae655f5e, reversing changes made to 1707e13b925f72543e67b12fd72ef75bd23a94f9. | 03 May 2019, 04:26:31 UTC |
| 895f40f | Mustafa M | 02 May 2019, 19:58:05 UTC | Consistently use the libblas constant (#31872) Also modify the docstring and attach to the module name. | 02 May 2019, 19:58:05 UTC |
| 6677406 | Mustafa M | 02 May 2019, 18:54:00 UTC | Remove unused download of 7z-extra | 02 May 2019, 18:54:00 UTC |
| b118e68 | Elliot Saba | 02 May 2019, 17:27:41 UTC | Upgrade and use NSIS for compression instead of extra 7z stage (#31879) Upgrade and use NSIS for compression instead of extra 7z stage | 02 May 2019, 17:27:41 UTC |
| 1707e13 | Keno Fischer | 02 May 2019, 12:16:50 UTC | Prevent stack overflow in Profile (#31893) As a follow up to #31693, this fixes the other place in Profile that recurses over Profile data, causing stack overflows. This should fix a bunch of the recent intermittent CI faults on linux32. | 02 May 2019, 12:16:50 UTC |
| 57ccd38 | Mus | 29 April 2019, 23:16:09 UTC | Update to latest NSIS and use NSIS for compression instead of extra 7z stage | 02 May 2019, 03:59:04 UTC |
| 5b637df | Chris Foster | 02 May 2019, 00:20:30 UTC | Ensure syntax error locations appear in backtraces (#31881) Errors which are thrown directly inside jl_toplevel_eval_flex do not get an entry in the backtrace. The most prominent of these are syntax errors, but there's other cases too. Fix this by constructing a julia expression throwing the appropriate error, and evaling it to generate a julia-level frame. | 02 May 2019, 00:20:30 UTC |
| 5eece3f | Jameson Nash | 01 May 2019, 16:57:30 UTC | apply: unify append_any implementations Rather than optimize two copies of the same function, we can just have one version of the function that handles everything. | 01 May 2019, 19:07:15 UTC |
| 56c496d | Jameson Nash | 01 May 2019, 16:43:57 UTC | inference: improve sizeof tfunc Primitive types also always have a sizeof. This is particularly relevant for `Ptr`, since `elsize` uses `sizeof(Ptr)` to compute the stride. | 01 May 2019, 19:07:07 UTC |
| 2aa5de5 | Katharine Hyatt | 01 May 2019, 15:15:56 UTC | Add suggested tests | 01 May 2019, 15:15:56 UTC |
| 7e1eb6a | Waldir Pimenta | 01 May 2019, 09:54:13 UTC | doc/build/windows: fix broken link & line breaks (#31884) | 01 May 2019, 09:54:12 UTC |
| 534c03b | Elliot Saba | 01 May 2019, 07:03:58 UTC | Disable building of tests for libunwind (#31886) Disable building of tests for libunwind | 01 May 2019, 07:03:58 UTC |
| e47df60 | Katharine Hyatt | 30 April 2019, 22:43:43 UTC | Three missing dense tests | 30 April 2019, 22:43:43 UTC |
| 2812a90 | Fredrik Ekre | 30 April 2019, 21:29:34 UTC | Disable building of tests for libunwind, which makes libunwind build with musl libc. See also https://github.com/JuliaPackaging/Yggdrasil/pull/13 | 30 April 2019, 22:09:09 UTC |
| 425c53d | Marco | 30 April 2019, 20:03:40 UTC | adding fast max and min (#31866) | 30 April 2019, 20:03:40 UTC |
| 8cc2f12 | Alex Arslan | 30 April 2019, 17:31:37 UTC | Stop exporting dot from LinearAlgebra.BLAS (#31878) LinearAlgebra exports the familiar `dot` function. The BLAS submodule has its own `dot` function, which it also exports, leading to problems if a user tries to access `dot` after `using` both LinearAlgebra and BLAS. Because this conflict currently exists, this change should be non-breaking, as it's difficult to resolve `dot` unqualified from BLAS before it's resolved from LinearAlgebra, so pretty much any consumer must already calling `BLAS.dot` as such. Fixes #31838. | 30 April 2019, 17:31:37 UTC |
| ffef661 | Viral B. Shah | 30 April 2019, 15:50:30 UTC | Move ARM to Tier 3 (#31880) Since we do not have ARM binaries at all on 1.1, it is best to move all ARM support to Tier 3. | 30 April 2019, 15:50:29 UTC |
| 5168e35 | Nicolau Werneck | 30 April 2019, 15:12:11 UTC | Recover meta nodes in replace_code_newstyle (#31871) Copy meta nodes from IRCode to CodeInfo | 30 April 2019, 15:12:11 UTC |
| 4700883 | Katharine Hyatt | 30 April 2019, 15:03:47 UTC | Missing undef array init tests (#31704) | 30 April 2019, 15:03:47 UTC |
| bb1cb52 | Viral B. Shah | 30 April 2019, 12:33:19 UTC | Merge branch 'master' of github.com:JuliaLang/julia | 30 April 2019, 12:33:19 UTC |
| 7ce6f5a | Viral B. Shah | 30 April 2019, 12:33:00 UTC | Fix broken link in README for required build tools. | 30 April 2019, 12:33:00 UTC |
| cce5a6b | Steven G. Johnson | 30 April 2019, 11:57:32 UTC | use Iterators.reverse for foldr etc. (#31781) | 30 April 2019, 11:57:31 UTC |
| 3d7e0d9 | Stefan Karpinski | 30 April 2019, 08:22:32 UTC | fix bad rebase in #23422 (#31874) | 30 April 2019, 08:22:32 UTC |
| 24ee806 | Elliot Saba | 30 April 2019, 00:23:06 UTC | Bump 7z version (#31873) Bump 7z version | 30 April 2019, 00:23:06 UTC |
| d40054b | Viral B. Shah | 29 April 2019, 23:18:37 UTC | Update the Editor section of README. | 29 April 2019, 23:18:37 UTC |
| c7fa684 | Viral B. Shah | 29 April 2019, 23:09:27 UTC | Add doc/build to source code organization in README. | 29 April 2019, 23:09:27 UTC |
| d562715 | Viral B. Shah | 29 April 2019, 22:35:04 UTC | Consistently name files in the doc/build directory. | 29 April 2019, 22:35:04 UTC |
| 19728c5 | Viral B. Shah | 29 April 2019, 22:32:14 UTC | Small cleanup for arm build documentation. | 29 April 2019, 22:32:14 UTC |
| a13c2cd | Viral B. Shah | 29 April 2019, 22:30:33 UTC | WIP: Refactoring of README.md (#31858) * Refactor the README to be simpler for the default build path. * Move detailed installation options and instructions to a `doc/build` | 29 April 2019, 22:30:33 UTC |
| a8c8377 | Jameson Nash | 24 April 2019, 18:29:07 UTC | types: lazy initialize the field-types when first needed For concrete datatypes, we still always initialize the fieldtypes immediately (so that we can immediately also compute the layout and other related properties). Note that this implies that constructing the fieldtype is no longer part of subtyping, (since construction no longer verifies these conditions), and so impossible constraints now lead to a computed value of `Union{}` being given for the type of that field, instead of throwing an error when trying to allocate the type. It might be possible to instead store this error, and rethrow it when someone tries to construct a (partially initialized) copy of the type and/or when the user (but not inference or introspection tools) tries to examine the fieldtype of the object. But I think this extra complexity (and additional failure cases to consider) isn't worthwhile to add. | 29 April 2019, 21:30:12 UTC |
| 0140ce8 | Rafael Fourquet | 29 April 2019, 20:43:39 UTC | add *(::Union{Regex, AbstractString, AbstractChar}...) (#23422) | 29 April 2019, 20:43:39 UTC |
| 458380c | Kevin Squire | 29 April 2019, 20:31:10 UTC | Loosen searchsorted* index type (fixes #30763, #31618) (#31633) * Might also address #31618 * Types of start and stop indicies are restricted to Integer and must be the same type * Note that this file is compiled early during bootstrap, and `one` is not yet available | 29 April 2019, 20:31:10 UTC |
| 826811b | Mus | 29 April 2019, 20:23:42 UTC | Bump 7z version | 29 April 2019, 20:23:42 UTC |
| 208d99c | Elliot Saba | 29 April 2019, 06:51:07 UTC | Merge pull request #31860 from JuliaLang/jn/tar-touch build: touch staged files after untar | 29 April 2019, 06:51:07 UTC |
| 6d02fe5 | Steven G. Johnson | 28 April 2019, 20:29:23 UTC | add findall(pattern, string) (#31834) | 28 April 2019, 20:29:22 UTC |
| a78f895 | Jeff Bezanson | 28 April 2019, 19:10:37 UTC | fix #31803, covariance handling in `fieldtype_tfunc` (#31848) | 28 April 2019, 19:10:37 UTC |
| 7d3d5e5 | Jameson Nash | 28 April 2019, 01:55:07 UTC | build: touch staged files after untar This avoids extracting mtime from the tar archives, improving incremental builds | 28 April 2019, 01:55:07 UTC |
| dc6c7c7 | Mustafa M | 27 April 2019, 03:49:15 UTC | Delete vagrant instructions (#31852) | 27 April 2019, 03:49:15 UTC |
| bacff35 | Keno Fischer | 27 April 2019, 02:13:26 UTC | Merge pull request #31790 from JuliaLang/kf/ambig Two fixes for bugs involving ambiguous signatures | 27 April 2019, 02:13:26 UTC |
| f3d17d3 | Keno Fischer | 22 April 2019, 00:52:27 UTC | Include ambiguous signatures when computing guard signatures After fixing #30118, the behavior of #31649 changed. Instead of crashing immediately upon attempting to print `spvec`, the first time would fail with the correct ambiguity. However, any subsequent attempts to print the same value would yet again result in a crash. Upon investigation, the series of events leading to this crash are as follows: 1. Showing the first error message attempts to put `convert(::Type{Any}, ::Nothing)` into the cache (matching the previous `convert(::Type{Any}, x::Any)` method). 2. jl_compilation_sig widens this to `convert(::DataType, x::Any)`. No guard signatures are allocated, because `ml_matches` does not return `convert(::Any, ::sparse_t)` because it is uncallable due to the ambiguity. 3. Upon the next call, we find `convert(::DataType, x::Any)` in the cache and use it. The subsequent successful return conflicts with type inference's result that this call will not return (i.e. is inferred to Union{}). To fix this change the `ml_matches` call in step 2 to include ambiguous matches, thus allocating guard entries for these signatures and preventing the cache match. That works for this issue, but I should note that I have seen additional crashes along similar lines while working on this fix, so there may be additional issues in the vicinity. Fixes #31649 | 26 April 2019, 22:01:56 UTC |
| a030b0a | Keno Fischer | 21 April 2019, 23:25:26 UTC | Don't inline methods that could potentially generate ambiguity errors If we have a call signature that's wide enough to cover one of the ambiguities of the methods we're inlining, refuse to do the inlining and require a dynamic dispatch instead. The other option would be to "union split" that signature out, but that would require a decent amount of extra code and it is not clear how important that speed up would be given that the signature is already going to be fairly wide if it intersects an ambiguity (so the dispatch gate might actually be better). If we encounter a case where that's useful, we can add that feature. For now, this fixes #30118. | 26 April 2019, 22:01:22 UTC |
| d71310e | Mustafa M | 26 April 2019, 19:51:29 UTC | Update uninstall julia heading level and a typo (#31846) | 26 April 2019, 19:51:29 UTC |
| 2d8c1ec | Viral B. Shah | 26 April 2019, 19:11:01 UTC | Remove circleci, old freebsd ci and vagrant (pkgeval) config files (#31841) * Remove circleci, old freebsd ci and vagrant (pkgeval) files * Remove vagrant from readme. * Move ack stuff to contrib/ack | 26 April 2019, 19:11:01 UTC |
