fc188a3 | Jeff Bezanson | 26 July 2018, 22:48:35 UTC | fix #28297, deserializing structs with compact Union-typed fields | 27 July 2018, 19:16:32 UTC |
e0cf3db | Jeff Bezanson | 26 July 2018, 21:17:00 UTC | fix #15722, support BigInts in precompiled modules and sysimg | 27 July 2018, 17:21:28 UTC |
b0075cb | Tim Besard | 26 July 2018, 08:37:55 UTC | Generate log message groups during macro expansion, if possible. | 27 July 2018, 17:15:42 UTC |
2df91c2 | Keno Fischer | 27 July 2018, 13:57:56 UTC | Fix left over deprecations in Pkg code (#28299) | 27 July 2018, 13:57:56 UTC |
ca7a692 | Kristoffer Carlsson | 27 July 2018, 11:19:10 UTC | split out TestHelpers into separate parts (#28292) | 27 July 2018, 11:19:10 UTC |
4b8c697 | Valentin Churavy | 26 July 2018, 22:05:10 UTC | bump to LLVM_BB rel 6.0.0-4 | 27 July 2018, 03:20:57 UTC |
df45146 | Keno Fischer | 26 July 2018, 03:06:50 UTC | Add LLVM patches for bugs introducing illegal ptrtoint The two patches are: rL323946 [LSR] Don't force bases of foldable formulae to the final type. D49832 [SCEV] Don't expand Wrap predicate using inttoptr in ni addrspaces | 27 July 2018, 03:20:57 UTC |
ec2f2e7 | Jarrett Revels | 26 July 2018, 23:04:32 UTC | refactor renumber_ir_elements to allow separate ssa and label changemaps (#28279) | 26 July 2018, 23:04:32 UTC |
38c8a03 | Jeff Bezanson | 26 July 2018, 04:38:36 UTC | fix #28277, replace `type` with `struct` in docs [ci skip] | 26 July 2018, 20:02:43 UTC |
2cb04e0 | Jeff Bezanson | 26 July 2018, 18:29:02 UTC | don't show module prefix in `:compact` printing mode (#27925) | 26 July 2018, 18:29:02 UTC |
bdfad24 | Alex Arslan | 26 July 2018, 16:48:56 UTC | Revert #27291 (force Git to use LF instead of CRLF) (#28280) Fixes #28241. | 26 July 2018, 16:48:56 UTC |
614d917 | Kristoffer Carlsson | 26 July 2018, 14:33:59 UTC | only freeze require world age in processess generating output (#28290) | 26 July 2018, 14:33:59 UTC |
1fa57f3 | Jeff Bezanson | 24 July 2018, 16:44:47 UTC | remove symbol references no longer used in the runtime system | 26 July 2018, 14:31:18 UTC |
04dbf7a | Jeff Bezanson | 30 June 2018, 21:02:49 UTC | improve specificity rule for typevars | 26 July 2018, 05:54:47 UTC |
cb33924 | Gaetano | 26 July 2018, 03:52:09 UTC | Update parallel-computing.md according to #19579 (#26527) * Update parallel computing documentation according to #19579 | 26 July 2018, 03:52:09 UTC |
937bd74 | Liozou | 23 July 2018, 21:45:02 UTC | Add optional arguments to @code_typed and @code_llvm | 26 July 2018, 03:06:16 UTC |
e7dd93b | Iblis Lin | 25 July 2018, 03:09:35 UTC | make allunique work for StepRange{Date,Day} | 26 July 2018, 03:01:32 UTC |
6f87a00 | Klaus Crusius | 24 July 2018, 12:56:09 UTC | reducedims for arrays with abstract element types. fix #28227 | 26 July 2018, 02:59:09 UTC |
9e7ae17 | Kristoffer Carlsson | 24 July 2018, 08:13:38 UTC | use a type parameter for stream in EachLine, improves performance | 26 July 2018, 02:56:17 UTC |
0f8c0bb | Steven G. Johnson | 26 July 2018, 02:09:18 UTC | support Unicode 11 via utf8proc 2.2 (#28266) | 26 July 2018, 02:09:18 UTC |
9fa4a22 | Jeff Bezanson | 25 July 2018, 02:06:01 UTC | fix #27907, inference not detecting some cycles during constant prop | 25 July 2018, 22:14:16 UTC |
c9bbcbd | Eric Davies | 25 July 2018, 22:13:43 UTC | Remove explicit message handling for easier log macro behaviour overriding (#28209) * Remove explicit message handling for easier log macro behaviour overriding * Put at-macrocall in TestHelpers * Add tests for no-message log macro MethodErrors | 25 July 2018, 22:13:43 UTC |
4969bdc | Jeff Bezanson | 24 July 2018, 20:56:02 UTC | still optimize top-level expressions with loops and `global` decls | 25 July 2018, 21:44:52 UTC |
df5a6b0 | Jeff Bezanson | 24 July 2018, 23:17:45 UTC | improve initial location info for compiled top-level exprs | 25 July 2018, 21:44:52 UTC |
1033b88 | Kristoffer Carlsson | 25 July 2018, 20:39:50 UTC | Revert "Enable PCRE UTF-8 validity string checks (#26731)" (#28259) This reverts commit 627173bcc986e4cd17a2e8d7ded7d977fbe68694. | 25 July 2018, 20:39:50 UTC |
68744d9 | Mathieu Besançon | 25 July 2018, 20:34:11 UTC | Doc base macros (#27949) | 25 July 2018, 20:34:11 UTC |
b189613 | Jeff Bezanson | 24 July 2018, 16:48:04 UTC | deprecate `:body` expression head This was used rarely and inconsistently, and is now redundant. | 25 July 2018, 20:07:31 UTC |
bd44a7a | Alex Arslan | 25 July 2018, 05:54:28 UTC | Reenable the test for #22566 on FreeBSD It was disabled in #24037 but reportedly seems to work okay when isolated into this separate, node 1 only test group. | 25 July 2018, 19:54:59 UTC |
3974369 | Alex Arslan | 25 July 2018, 03:25:02 UTC | Move tests for #13559, #22566, and jl_exit_on_sigint to stress.jl | 25 July 2018, 19:54:59 UTC |
20fc0f5 | Alex Arslan | 24 July 2018, 23:09:49 UTC | Separate stress tests out and run them on node 1 This ensures that they don't interfere with other tests. | 25 July 2018, 19:54:59 UTC |
6476f51 | Valentin Churavy | 25 July 2018, 18:17:40 UTC | Merge pull request #28275 from JuliaLang/tb/jiteventlistener Fix JIT event listener registration. | 25 July 2018, 18:17:40 UTC |
2864093 | Valentin Churavy | 25 July 2018, 15:13:15 UTC | Merge pull request #28273 from DokFaust/df/fixD44892patch Fixes an undef compilation warning | 25 July 2018, 15:13:15 UTC |
adc2d92 | Jeff Bezanson | 23 July 2018, 22:49:26 UTC | fix #27529, generator exprs should turn off space-sensitive after `for` | 25 July 2018, 14:56:47 UTC |
d3d603f | Tim Besard | 25 July 2018, 14:54:14 UTC | Fix JIT event listener registration. | 25 July 2018, 14:54:14 UTC |
930d110 | Jameson Nash | 25 July 2018, 14:30:31 UTC | Merge pull request #28264 from JuliaLang/jn/irshow2 Further refinements to printing IR | 25 July 2018, 14:30:31 UTC |
7d7c60c | DokFaust | 25 July 2018, 10:55:42 UTC | Fixes an undef compilation warning | 25 July 2018, 10:55:42 UTC |
77342e2 | Jeff Bezanson | 24 July 2018, 15:39:45 UTC | fix #28256, subtyping bug where `simple_meet` failed to check egal | 25 July 2018, 02:11:51 UTC |
77a4cb5 | Rafael Fourquet | 24 July 2018, 21:03:27 UTC | LineEdit: revert accidental rename of endofline to lastindexline (#28257) Was done in the endof -> lastdindex rename (742a5399). | 24 July 2018, 21:03:27 UTC |
aa44d01 | Jeff Bezanson | 24 July 2018, 18:46:02 UTC | fix #28244, segfault on `@macroexpand` with macro returning an invalid expr (#28247) | 24 July 2018, 18:46:02 UTC |
03f079d | Steven G. Johnson | 24 July 2018, 18:44:39 UTC | allow 24:00 DateTime (#28239) fixes #28203 | 24 July 2018, 18:44:39 UTC |
bb443fe | Fengyang Wang | 24 July 2018, 18:40:06 UTC | Consider newlines as spaces for space-sensitive parsing (#20265) | 24 July 2018, 18:40:06 UTC |
384c542 | Iblis Lin | 24 July 2018, 17:43:09 UTC | minor devdocs update (#28233) update link of jl_parse_opts | 24 July 2018, 17:43:09 UTC |
68fe358 | Jameson Nash | 24 July 2018, 17:07:57 UTC | ssair-verify: send output to the correct IO stream | 24 July 2018, 17:25:49 UTC |
20c44fc | Jameson Nash | 24 July 2018, 15:00:06 UTC | show-ir: print the basic-block label when possible When basic-block information is available, it's preferable that we print that information when displaying code IR, as it is more readable (and more consistent between CodeInfo and IRCode form). Warning: if the basic blocks are not computed correctly, this might print misleading information. However, this seems unlikely, since there should be a basic-block start recorded at every label used as a destination. | 24 July 2018, 17:25:49 UTC |
fc3a709 | Jameson Nash | 24 July 2018, 13:36:36 UTC | Revert "Revert "correct AST show regressions, fix formatting around pending nodes"" This reverts commit b516e81c75c33f09a2e79e0a8a42d3057700bc24. | 24 July 2018, 17:25:49 UTC |
c5a8760 | Jameson Nash | 24 July 2018, 13:34:04 UTC | Revert "IRCode: fix printing of pending node basic block" This reverts commit 4d20a9f9190490b9cafec3801a8e7d5689f8becb. | 24 July 2018, 16:20:10 UTC |
978577f | Keno Fischer | 24 July 2018, 14:04:53 UTC | Allow .V to be null in emit_unionmove (#28248) We generally handle the case where the .V of a split union is as small as the active element of the union. If the active element happens to be a ghost is thus seems reasonable for the split union to not have any storage at all. Such a union is generated e.g. if we widen from an all-ghost split union to one that includes sized elements, e.g. in the following example: ``` function foo() x = (1, 2) if rand() < 0.5 if rand() < 0.5 y = nothing else y = missing end x = y end x end ``` Fixes #28208 | 24 July 2018, 14:04:53 UTC |
69014c8 | Klaus Crusius | 24 July 2018, 13:56:51 UTC | added missing method for `Profile.callers` (#28225) added missing method for `Profile.callers` | 24 July 2018, 13:56:51 UTC |
cbb6433 | Stefan Karpinski | 24 July 2018, 04:22:41 UTC | load path: deduplicate harder (#28249) It was previously fairly easy to end up with duplicate entries in `LOAD_PATH` when controlled by `JULIA_LOAD_PATH`; this eliminates duplicates when setting `LOAD_PATH` from `JULIA_LOAD_PATH` on startup. There's no functional change since only the first instance of the same entry in the load path matters. | 24 July 2018, 04:22:41 UTC |
789ad04 | Keno Fischer | 24 July 2018, 03:59:36 UTC | Add special case in verification for trailing statements after :enter (#28246) In #28230, I had the compiler make use of some not-quite-legal IR to quickfix #28224. Somewhat ironically, a simultaneous PR I was working on made the verifier realize that this IR is illegal and complain during bootstrap. Add a special case in verification to allow this pattern for now. | 24 July 2018, 03:59:36 UTC |
9b88e46 | Jacob Quinn | 24 July 2018, 03:28:25 UTC | Merge pull request #28243 from JuliaLang/jq/isbitsunionarrayfixes Fix issue where isbits Union arrays didn't copyto or vcat correctly b… | 24 July 2018, 03:28:25 UTC |
97fb554 | Simon Byrne | 23 July 2018, 18:15:43 UTC | expand binomial docs | 23 July 2018, 23:24:51 UTC |
ce9bc6c | Simon Byrne | 23 July 2018, 21:37:24 UTC | handle round(x; base=...) (#28199) Fixes #28160. | 23 July 2018, 21:37:24 UTC |
e6f5098 | Steven G. Johnson | 23 July 2018, 21:13:13 UTC | allow re-importing the same module from different owners (#28197) * allow re-importing the same module from different owners * test | 23 July 2018, 21:13:13 UTC |
89285cb | quinnj | 23 July 2018, 20:43:28 UTC | Fix issue where isbits Union arrays didn't copyto or vcat correctly because of new layout and presence of internal array padding | 23 July 2018, 20:43:28 UTC |
10002fc | Fredrik Ekre | 23 July 2018, 19:53:50 UTC | Revert part of fbc30cd6c744c8cef82cb6fd54d10bdce9b36692 (#28206) don't load Pkg automatically in interactive use | 23 July 2018, 19:53:50 UTC |
88a8c86 | Carlo Baldassi | 23 July 2018, 11:10:57 UTC | Merge pull request #28232 from JuliaLang/cb/copytofix Fix error message in copyto!(::AbstractArray, ::Any) | 23 July 2018, 11:10:57 UTC |
fdad505 | Keno Fischer | 22 July 2018, 20:29:02 UTC | Avoid printing to stdout in show test | 23 July 2018, 04:14:44 UTC |
4d20a9f | Jameson Nash | 19 July 2018, 17:54:34 UTC | IRCode: fix printing of pending node basic block fix #28173 | 23 July 2018, 04:14:44 UTC |
79f8a28 | Jameson Nash | 17 July 2018, 22:13:38 UTC | show: fix GlobalRef printing for quoted operator-identifiers fix #27966 | 23 July 2018, 04:14:44 UTC |
a9b22eb | Jacob Quinn | 22 July 2018, 23:07:20 UTC | Merge pull request #27924 from JuliaLang/jq/isbitsunionarrays Improve performance of resizing isbits Union Arrays | 22 July 2018, 23:07:20 UTC |
1e5ec30 | Keno Fischer | 22 July 2018, 19:29:40 UTC | Quick fix for Upsilon node intersecting loop backedge The issue is well illustrated by the MWE in #28224: ``` julia> @noinline make_error(n) = n == 5 ? error() : true make_error (generic function with 1 method) julia> function foo() z = 0 try while make_error(z) z+=1 end catch end return z end foo (generic function with 1 method) julia> foo() 0 ``` In the typed code, we see that the initial Upsilon node, marking the value at the beginning of the critical region, is accidentally put after the loop backedge: ``` 1 ─ nothing 3 2 ─ :($(Expr(:enter, 4))) 4 3 ┄ %3 = φ (2 => 0, 3 => %6)::Int64 │ %4 = ϒ (0)::Int64 │ invoke Main.make_error(%3::Int64) 5 │ %6 = Base.add_int(%3, 1)::Int64 │ %7 = ϒ (%6)::Int64 └── goto 3 4 ┄ %9 = φᶜ (%4, %7)::Int64 └── :($(Expr(:leave, 1))) 8 5 ─ return %9 6 ─ goto 4 ``` Thus getting reset on every loop iteration. This is a quick fix to instead insert the initial upsilon node in the :enter's BB. That's not technically supposed to work, since the :enter is a terminator, but it works fine for now, since we never look at terminators again after the initial CFG construction. Once we round trip through the statement-addressed representation, the CFG is correct. In the future, compaction should fix up the CFG during insertion, but we currently don't touch the CFG at all during optimization, so we can't do that at the moment. This solution should hold us over until then. Fixes #28224 | 22 July 2018, 22:17:41 UTC |
886308d | Carlo Baldassi | 22 July 2018, 21:34:35 UTC | Fix error message in copyto!(::AbstractArray, ::Any) | 22 July 2018, 21:34:35 UTC |
41e749f | Mauro | 22 July 2018, 18:03:45 UTC | WeakKeyDict with no conversion on key-insertion (#24941) (#28198) | 22 July 2018, 18:03:45 UTC |
04db1bd | Jeff Bezanson | 20 July 2018, 21:26:39 UTC | add some fast paths to subtyping to help loading ApproxFun | 22 July 2018, 17:19:53 UTC |
5433460 | Jeff Bezanson | 17 July 2018, 23:04:25 UTC | improve variable promotion in closure conversion This helps #15276 and #19668. Now more variables that dominate all uses can be identified in the front-end and un-`Box`ed. | 22 July 2018, 16:35:05 UTC |
39eb68b | Alex Arslan | 22 July 2018, 16:28:50 UTC | Add `missing` support in `isapprox` (#28094) * Add `missing` support in `isapprox` `isapprox` now behaves the same way as `==` in the presence of missing values: if either argument is `missing`, the result is `missing`. Fixes #28088. * Define `norm` for missing values | 22 July 2018, 16:28:50 UTC |
03f0ae5 | David Anthoff | 28 May 2018, 18:14:08 UTC | Configure git to always use LF line endings in the working directory | 22 July 2018, 15:51:29 UTC |
f5d8f98 | quinnj | 22 July 2018, 05:08:52 UTC | Fix issue in jl_array_shrink where the offset wasn't correctly multiplied by a->elsize, so the array's data pointer got messed up. | 22 July 2018, 05:30:52 UTC |
545e1d3 | quinnj | 22 July 2018, 05:07:08 UTC | Refactor isbits Union arrays to treat the type tag bytes like a proper jl_array_t* with potential a->offset and a->maxsize values. Before, any resizing operations on isbits Union arrays (push, pushfirst, insert, append, deleteat, slice, etc.) incurred extra cost by having to constantly move the type tag bytes to be directly after the last array element. By treating them rather as jl_array_t* objects, sharing the a->offset and a->maxsize fields with the parent array, resizing operations now only require moving bytes around if the parent array's data itself must be moved. Fixes #27825 and #27809. | 22 July 2018, 05:30:42 UTC |
11cba0e | Keno Fischer | 20 July 2018, 19:54:08 UTC | Disallow double edges in IR A double edge is an edge of the form: ``` 4 - goto 5 if not %%x 5 - ... ``` i.e. where both the fall through and the explicit edge of the conditional branch go to the same basic block (note however that such branches are basically no-ops). LLVM allows this (with some restrictions - e.g. if there is a phi node the value on both edges has to be the same), so I also decided to allow it. However, it turns out to cause an enormous amount of pain because edges are no longer uniquely identified by just the two basic blocks at the edge's end points. In the near future, I want to start doing CFG transforms at the IR level and these double edges require a special case for basically every operation. Thus, in preparation for that, let's disallow such edges as an IR invariant. The trade off here is that when we modify conditional branches now, we have to check whether they would be a noop and delete them in that case. However, a) we likely want to do that anyway for IR compactness and b) it seems a lot less fidegedy than dealing with double edge. | 21 July 2018, 23:38:00 UTC |
b516e81 | Keno Fischer | 21 July 2018, 18:19:20 UTC | Revert "correct AST show regressions, fix formatting around pending nodes" | 21 July 2018, 20:10:03 UTC |
cb36653 | Morten Piibeleht | 21 July 2018, 18:48:51 UTC | Bump Documenter to 0.19.1 (#28220) | 21 July 2018, 18:48:51 UTC |
49577e9 | Jameson Nash | 21 July 2018, 12:14:17 UTC | Merge pull request #28195 from JuliaLang/jn/irshow correct AST show regressions, fix formatting around pending nodes | 21 July 2018, 12:14:17 UTC |
93d89c5 | Jameson Nash | 20 July 2018, 04:30:35 UTC | ir-show: more improvements print types and statement numbers in more cases where they are useful | 21 July 2018, 03:33:17 UTC |
7bb5600 | Jameson Nash | 19 July 2018, 17:54:34 UTC | IRCode: fix printing of pending node basic block fix #28173 | 21 July 2018, 03:33:17 UTC |
aa43070 | Jameson Nash | 17 July 2018, 22:13:38 UTC | show: fix GlobalRef printing for quoted operator-identifiers fix #27966 | 21 July 2018, 03:33:17 UTC |
6cbb52b | Jameson Nash | 17 July 2018, 21:52:43 UTC | show: reduce code duplication in printing IR code fix #27644 fix #27965 (test coverage provided now by using existing methods for showing IR) | 21 July 2018, 03:33:17 UTC |
1d3c9c8 | Keno Fischer | 20 July 2018, 21:32:18 UTC | Correct assertion checking for deleted typemap entry This assertion was added in #27568. However from my understanding the first part of the assertion is wrong, since that PR added logic to mark deleted typemap entries as min_world 0, max_world 1. It seems valid for a typemap entry to be deleted, even if the referenced (while the reverse does not seem valid). Fixes #28213 | 21 July 2018, 02:46:45 UTC |
383751c | Jeff Bezanson | 20 July 2018, 20:02:44 UTC | fix #28159, better field reflection functions for NamedTuple (#28200) | 20 July 2018, 20:02:44 UTC |
f5c46c8 | Simon Byrne | 22 May 2018, 03:49:13 UTC | simplify cbrt codepaths | 20 July 2018, 20:00:30 UTC |
ff0de87 | Patrick Kofod Mogensen | 08 February 2018, 11:00:14 UTC | Add Julia port of cbrt from Openlibm. | 20 July 2018, 20:00:30 UTC |
e7acea7 | Kristoffer Carlsson | 20 July 2018, 13:30:20 UTC | update Documenter and fix build path for pdf (#28207) | 20 July 2018, 13:30:20 UTC |
9ed6287 | Jameson Nash | 20 July 2018, 04:01:05 UTC | performance: mark REPL and Doc code as non-specializeable (#28065) Implicit return is bad for compiler performance (and sometimes runtime performance) and can adversely affect code readability, so every function which does _not_ return a value should end in a `return` statement. Here, we also introduce a new meaning to the `@nospecialize` macro, and a new macro `@specialize` to reverse its effect. When used without arguments, it applies to all arguments of the parent. In local scope, this means the containing function. In global scope, this means all methods subsequently defined. | 20 July 2018, 04:01:05 UTC |
21be910 | Steven G. Johnson | 19 July 2018, 15:04:22 UTC | fix unicode shift-selection in REPL (fixes #28183) | 20 July 2018, 02:24:49 UTC |
cc7d9dd | Bogumił Kamiński | 19 July 2018, 18:30:21 UTC | improve speed of string and repeat for SubString{String} and make string(::SubString{String}) return String | 20 July 2018, 02:21:10 UTC |
7ff46a5 | Ed Schmerling | 20 July 2018, 02:17:56 UTC | Fix @which for getindex methods using [] syntax (#28122) | 20 July 2018, 02:17:56 UTC |
67b06af | Matt Bauman | 13 July 2018, 19:11:23 UTC | Fix type stability of Broadcast.flatten Fixes #27988. | 20 July 2018, 02:15:28 UTC |
5d876f4 | Kristoffer Carlsson | 20 July 2018, 00:21:33 UTC | allow packages to find UUID in project not in deps (#27932) but print a warning that this is happening | 20 July 2018, 00:21:33 UTC |
9e9c4f8 | Jeff Bezanson | 19 July 2018, 16:20:43 UTC | fix inlining cost of IntrinsicFunctions when consts have been widened along with the llvmcall fix, this fixes #28078 | 19 July 2018, 21:54:34 UTC |
2c3ef2a | Jeff Bezanson | 18 July 2018, 18:00:02 UTC | update llvmcall for new IR. fixes #27694 | 19 July 2018, 21:54:34 UTC |
c6a949a | Jeff Bezanson | 19 July 2018, 18:28:19 UTC | add type parameter to `AbstractChannel`. fixes #28124 (#28174) | 19 July 2018, 18:28:19 UTC |
b012ab3 | Valentin Churavy | 19 July 2018, 14:45:21 UTC | Merge pull request #27466 from DokFaust/df/operflist Reintegrate external profiling support | 19 July 2018, 14:45:21 UTC |
91c2da8 | Tanmay Mohapatra | 19 July 2018, 07:13:34 UTC | accept AbstractString for BigFloat construction (#28169) This is a minor change to allow `BigFloat` construction from `AbstractString`. E.g.: ``` julia> BigFloat(strip(" 42")) 4.2e+01 ``` Without this: ``` julia> BigFloat(strip(" 42")) ERROR: MethodError: Cannot `convert` an object of type SubString{String} to an object of type BigFloat ... ``` | 19 July 2018, 07:13:34 UTC |
68366d6 | Kristoffer Carlsson | 19 July 2018, 06:38:23 UTC | use inbounds in generic matvecmul (#28163) | 19 July 2018, 06:38:23 UTC |
8d0fcdb | DokFaust | 29 May 2018, 15:08:41 UTC | Added the capability of looking up JIT symbols for perf Moved Oprofile Wrapper, implemented RegisterJITEventListener and NotifyFinalized Added PerfJITEventListener as an LLVM Patch Added IntelJITEventListener Support Added LLVM Patch, removed redundant Oprofile code, wipe blanklines, Register OProfile JITEvent, fixed oprofile compile flag, rebased llvm 6.0 checksums Added Perf profiling support as an LLVM patch. | 19 July 2018, 04:30:27 UTC |
6c47c24 | Kristoffer Carlsson | 18 July 2018, 21:17:54 UTC | make simd loop over ranges perform better (#28166) | 18 July 2018, 21:17:54 UTC |
a5ee0cc | Steven G. Johnson | 18 July 2018, 14:39:51 UTC | don't allow U+0387 in identifiers | 18 July 2018, 19:34:54 UTC |
76452ef | Jeff Bezanson | 17 July 2018, 18:28:57 UTC | fix bug in top-level expressions in --compile=min mode | 18 July 2018, 19:29:42 UTC |
f000f14 | Kristoffer Carlsson | 18 July 2018, 19:08:07 UTC | add note to FAQ how to check if a file is executed using julia file.jl (#28137) * Update faq.md | 18 July 2018, 19:08:07 UTC |