5c19d44 | Keno Fischer | 22 December 2017, 19:27:24 UTC | Iterate WIP | 22 December 2017, 19:27:24 UTC |
0e66b1b | Keno Fischer | 22 December 2017, 14:38:06 UTC | indexed_next -> indexed_iterate | 22 December 2017, 14:38:06 UTC |
6b8f83b | Keno Fischer | 22 December 2017, 12:37:18 UTC | Import Jameson's lowering changes | 22 December 2017, 12:37:18 UTC |
915b2e8 | Steven G. Johnson | 21 December 2017, 21:41:59 UTC | fix spawn to work with AbstractPipe (#25209) | 21 December 2017, 21:41:59 UTC |
1e0d26b | Chris Foster | 21 December 2017, 20:45:02 UTC | Rename unshift!/shift! to pushfirst!/popfirst! (#25100) | 21 December 2017, 20:45:02 UTC |
8872f90 | Milan Bouchet-Valat | 21 December 2017, 19:54:03 UTC | Make arrays and ranges hash and compare equal (#16401) * Make vectors and ranges hash and compare equal When hashing AbstractVectors, first check whether their first elements are equal to a range, and hash them as a such if that's the case. This allows for O(1) hashing of (some) ranges consistent with AbstractArrays, which means they can now compare equal. Types which have a regular range step have to use the new TypeRangeStep trait to enable O(1) hashing rather than the O(N) AbstractArray fallback. Apply the new trait to date ranges which have a regular step. Add tests for the new behaviors. | 21 December 2017, 19:54:03 UTC |
bd04c13 | Dream Scatter | 21 December 2017, 14:37:11 UTC | add ⟂ to infix operator precedence (#24404) | 21 December 2017, 14:37:11 UTC |
4705ba9 | Jeff Bezanson | 21 December 2017, 13:47:10 UTC | Merge pull request #23273 from JuliaLang/jb/ctors deprecate fallback constructor | 21 December 2017, 13:47:10 UTC |
5fcd0d9 | Tim Holy | 21 December 2017, 12:46:09 UTC | Revert "Merge pull request #25219 from Sacha0/higho" This reverts commit fa856bb74b899d510f0746609d27485d2e5c561f, reversing changes made to f9e2d99285fd4c3ea038e740ef82307dcb5ed598. | 21 December 2017, 12:46:09 UTC |
d24ccd4 | Valentin Churavy | 21 December 2017, 11:15:24 UTC | Merge pull request #25207 from JuliaLang/sk/xreload remove reload | 21 December 2017, 11:15:24 UTC |
fa856bb | Tim Holy | 21 December 2017, 08:44:19 UTC | Merge pull request #25219 from Sacha0/higho optimize and fix map/broadcast over Adjoint/Transpose vectors | 21 December 2017, 08:44:19 UTC |
f9e2d99 | Stefan Karpinski | 21 December 2017, 04:43:09 UTC | Merge pull request #25158 from c42f/depwarn-caller-fix Depwarn caller fix | 21 December 2017, 04:43:09 UTC |
052d9ab | Sacha Verweij | 21 December 2017, 04:26:49 UTC | Eliminate temporary in map/broadcast over Adjoint/Transpose vectors, and fix inference. | 21 December 2017, 04:26:49 UTC |
b57a592 | Jameson Nash | 21 December 2017, 03:31:37 UTC | make Ref(x) always construct a RefValue(x) object (#21527) the old behavior is now directly a feature of `convert` only instead | 21 December 2017, 03:31:37 UTC |
31006cf | Jeff Bezanson | 07 September 2017, 22:40:53 UTC | deprecate convert-to-construct fallback | 21 December 2017, 00:09:49 UTC |
d55148e | Jeff Bezanson | 30 August 2017, 21:14:01 UTC | consolidate `Associative` conversions, and allow `AbstractSet` conversions | 21 December 2017, 00:09:06 UTC |
25f26e1 | Jeff Bezanson | 30 August 2017, 18:01:28 UTC | update manual to explain the new mental model of `convert` vs. construct | 21 December 2017, 00:09:05 UTC |
0e726e6 | Jeff Bezanson | 22 August 2017, 18:22:52 UTC | change more `convert` methods to constructors deprecate a few more sketchy `convert` methods | 21 December 2017, 00:09:05 UTC |
97e7c0c | Jeff Bezanson | 15 August 2017, 20:00:41 UTC | remove fallback constructor from Core.Inference Instead, define explicit constructors for built-in integer types in boot.jl. | 21 December 2017, 00:06:58 UTC |
555264e | Stefan Karpinski | 20 December 2017, 23:41:26 UTC | Merge pull request #25162 from JuliaLang/sk/voider a more earnest attempt at renaming Void | 20 December 2017, 23:41:26 UTC |
1a6462e | Stefan Karpinski | 20 December 2017, 23:35:34 UTC | fix nextind bug for invalid UTF-8 (#25214) reported by bkamins in #24420 | 20 December 2017, 23:35:34 UTC |
0f4d5db | Stefan Karpinski | 18 December 2017, 18:06:27 UTC | NEWS and deprecation for Void => Nothing rename | 20 December 2017, 21:57:31 UTC |
e5fa92b | Stefan Karpinski | 18 December 2017, 15:16:13 UTC | rename Void => Cvoid, Nothing (depending on usage) git ls-files base doc examples src stdlib test | xargs perl -i -ple '/\b(ccall|cglobal|cfunction)\b/ && s/\bVoid\b/Cvoid/g' git ls-files base doc examples src stdlib test | xargs perl -i -ple 's/\bPtr\{Void\}/Ptr{Cvoid}/g' git ls-files base doc examples src stdlib test | xargs perl -i -ple 's/\Ref\{Void\}/Ref{Cvoid}/g' perl -i -ple 's/\bVoid\b/Cvoid/g' base/atomics.jl base/libgit2/status.jl base/random/dSFMT.jl doc/src/manual/calling-c-and-fortran-code.md test/llvmcall.jl test/threads.jl git ls-files base doc examples src stdlib test | xargs perl -i -ple 's/\bVoid\b/Nothing/g' | 20 December 2017, 21:57:30 UTC |
84ce70b | Stefan Karpinski | 15 December 2017, 05:51:35 UTC | const Cvoid = Void | 20 December 2017, 21:54:56 UTC |
617afff | Jonathan D Johnston | 20 December 2017, 17:20:12 UTC | fix #25137, return value of `dlclose` on windows (#25175) - Invert boolean value returned by Windows FreeLibrary so it matches that of POSIX dlclose - Add test of jl_dlclose/Libdl.dlclose to test/libdl.jl | 20 December 2017, 17:20:12 UTC |
564ecb0 | Tim Holy | 20 December 2017, 11:49:11 UTC | Merge pull request #25050 from JuliaLang/teh/delete_method Support method deletion | 20 December 2017, 11:49:11 UTC |
23362c0 | Stefan Karpinski | 20 December 2017, 09:46:45 UTC | remove reload | 20 December 2017, 10:45:09 UTC |
7b1c06a | Rafael Fourquet | 20 December 2017, 09:43:40 UTC | Merge pull request #23528 from JuliaLang/rf/few-set-methods add few missing methods for set-like operations | 20 December 2017, 09:43:40 UTC |
eb0d9e4 | Katharine Hyatt | 20 December 2017, 09:42:05 UTC | Some small grammar nits (#25203) | 20 December 2017, 09:42:05 UTC |
6e2f503 | Jeff Bezanson | 20 December 2017, 06:50:26 UTC | fix #25055, `quote end` misoptimized to return same object (#25199) | 20 December 2017, 06:50:26 UTC |
b6d2731 | Katharine Hyatt | 20 December 2017, 05:23:35 UTC | Merge pull request #25191 from JuliaLang/ksh/evenmorearrays Xrefs and more examples for array indexing | 20 December 2017, 05:23:35 UTC |
2d45ef5 | Jeff Bezanson | 20 December 2017, 04:02:31 UTC | preserve `do` syntax in AST. part of #21774 (#23718) | 20 December 2017, 04:02:31 UTC |
b537f78 | Jameson Nash | 20 December 2017, 01:47:40 UTC | Merge pull request #25181 from JuliaLang/jn/disasm-symbolicator disasm: improve symbolicator | 20 December 2017, 01:47:40 UTC |
1319567 | Sacha Verweij | 19 December 2017, 22:20:16 UTC | Merge pull request #25148 from Sacha0/despecial update context-independent and remove context-dependent lowering of ' | 19 December 2017, 22:20:16 UTC |
6bf6973 | Rafael Fourquet | 02 September 2017, 12:20:16 UTC | add mutating set-like operations for AbstractVector | 19 December 2017, 21:43:07 UTC |
ea3639e | Stefan Karpinski | 19 December 2017, 20:54:58 UTC | Make inbounds macros expression-like (#25033) | 19 December 2017, 20:54:58 UTC |
3c22019 | Jameson Nash | 19 December 2017, 05:30:51 UTC | disasm: improve symbolicators ability to handle constants | 19 December 2017, 18:17:22 UTC |
7a89d43 | kshyatt | 19 December 2017, 17:16:40 UTC | Xrefs and more examples for array indexing | 19 December 2017, 17:16:40 UTC |
de8774a | Rafael Fourquet | 01 September 2017, 13:31:42 UTC | fix perf regression in filter/filter! | 19 December 2017, 12:34:00 UTC |
35d280d | Rafael Fourquet | 08 August 2017, 18:48:18 UTC | add few missing methods for set-like operations | 19 December 2017, 12:34:00 UTC |
c16c0cb | Rafael Fourquet | 19 December 2017, 11:51:50 UTC | Merge pull request #25029 from JuliaLang/rf/intset-offset Make BitSet store any Int | 19 December 2017, 11:51:50 UTC |
55852ee | Rafael Fourquet | 14 December 2017, 11:06:59 UTC | add fuzzy testing for some Set and BitSet functions | 19 December 2017, 08:46:43 UTC |
a4371b2 | Rafael Fourquet | 12 December 2017, 10:41:50 UTC | BitSet: add a couple small optimizations * isempty: using _check0 instead of all makes it 35% faster * ==: checking first non-overlapping parts is more likely to be faster, as the the lower and upper parts of the bits field are unlikely to be zero (at least for a freshly created BitSet) | 19 December 2017, 08:45:12 UTC |
55968e7 | Rafael Fourquet | 12 December 2017, 10:36:47 UTC | BitSet: use _div64 & _mod64 from bitarray.jl | 19 December 2017, 08:45:12 UTC |
dd6a94a | Rafael Fourquet | 12 December 2017, 10:33:00 UTC | re-implement rand(::BitSet) | 19 December 2017, 08:45:12 UTC |
7f44009 | Rafael Fourquet | 11 December 2017, 17:24:10 UTC | optimize updating a value in BitSet _setint! was delegating to the bitarray code, but then some work was being redone in get_chunks_id. So we split this functionality into lower level functions. | 19 December 2017, 08:45:12 UTC |
13cdaeb | Rafael Fourquet | 11 December 2017, 16:51:27 UTC | BitSet push! & pop! : don't check for emptyness every time When a BitSet is empty, its offset must be initialized. But checking for emptyness in each invocation of _setint! was costly, so we put the check in a branch which is called less often. | 19 December 2017, 08:45:12 UTC |
ed692cd | Rafael Fourquet | 11 December 2017, 14:37:21 UTC | BitSet: use 0-based bit-indexing The 1-based indexing was forcing to promote Int to Int128 in chk_indice & chk_offset, to allow correct computation with extreme values (e.g. BitSet([typemin(Int)])). 1-based indexing was a left-over from the former BitVector as underlying implementation. Switching to 0-based indexing allows both a more-direct mapping between the model and the implementation, and better performance. | 19 December 2017, 08:45:12 UTC |
cb49741 | Rafael Fourquet | 02 August 2017, 15:24:15 UTC | morph BitSet into a real Int Set Add an offset field to BitSet acting as an infimum of stored values, which can be adjusted on-the-fly. BitSet then becomes able to store negative integers, and doesn't have to have elements centered around 0. | 19 December 2017, 08:45:08 UTC |
e48eaaf | Katharine Hyatt | 19 December 2017, 07:35:04 UTC | A few more references for arrays manual (#25173) | 19 December 2017, 07:35:04 UTC |
813ad93 | Fredrik Ekre | 19 December 2017, 07:24:53 UTC | emit a friendly error message for workspace(), close #25166 (#25170) | 19 December 2017, 07:24:53 UTC |
3216445 | Fredrik Ekre | 19 December 2017, 07:24:06 UTC | fill[stored]! for Symmetric and Hermitian matrices (#25152) | 19 December 2017, 07:24:06 UTC |
d5d12ad | Rafael Fourquet | 10 December 2017, 14:42:17 UTC | BitSet: swap the underlying BitVector for Vector{UInt64} The BitVector had already started to show its limits for BitSet needs. Juggling at 3 layers with BitSet, BitVector, and Vector{UInt64} was becoming confusing at times. Updating this code to handle negative integers only increased the confusion. Some modest performance improvements (~20%) could be achieved in the process for some functions. | 19 December 2017, 07:01:57 UTC |
e704d22 | Jameson Nash | 19 December 2017, 00:13:46 UTC | disasm: remove invalid OpInfoLookup code This lets LLVM fall through to the Symbol symbolicator, where we already have the right answer Otherwise, this code might get in the way, if the instruction encoding was not just the literal pointer. | 19 December 2017, 05:32:29 UTC |
1044335 | Katharine Hyatt | 19 December 2017, 01:41:25 UTC | More doctests for array fun (#24920) * Simple cleanup and doctest for views macro * Doctests for accumulate! and copy! | 19 December 2017, 01:41:25 UTC |
6e0549e | Sacha Verweij | 19 December 2017, 01:10:41 UTC | Merge pull request #25150 from Sacha0/history move 0.6 NEWS.md section to HISTORY.md | 19 December 2017, 01:10:41 UTC |
6e3c1ce | Sacha Verweij | 18 December 2017, 20:44:31 UTC | Remove special lowering of ' in *, /, and \ expressions to A[c]_(mul|ldiv|rdiv)_B[c]. | 19 December 2017, 01:09:03 UTC |
664fc98 | Sacha Verweij | 17 December 2017, 22:48:01 UTC | Make A' lower to Core.postfixapostrophize(A) with default def = Adjoint(A) and clean up fallout. | 19 December 2017, 01:09:03 UTC |
07740b6 | Jameson Nash | 18 December 2017, 22:50:03 UTC | add tests for getfield-overloading (#25164) tests for PR #24960 | 18 December 2017, 22:50:03 UTC |
90cf0ba | Sacha Verweij | 17 December 2017, 01:56:37 UTC | Rewrite isolated ' calls in test/ to preserve behavior through ' lowering changes. | 18 December 2017, 21:38:32 UTC |
9ec817f | Sacha Verweij | 16 December 2017, 22:16:26 UTC | Rewrite isolated ' calls in stdlib/ to preserve behavior through ' lowering changes. | 18 December 2017, 21:38:31 UTC |
8699d98 | Sacha Verweij | 16 December 2017, 22:03:07 UTC | Rewrite isolated ' calls in base/ to preserve behavior through ' lowering changes. | 18 December 2017, 21:38:31 UTC |
5abe9b1 | Stefan Karpinski | 18 December 2017, 19:26:13 UTC | define peek(::TTYTerminal), fixes #25098 (#25161) | 18 December 2017, 19:26:13 UTC |
0272186 | Sacha Verweij | 18 December 2017, 19:02:06 UTC | Merge pull request #25125 from Sacha0/depdotap remove special lowering for and deprecate .' | 18 December 2017, 19:02:06 UTC |
ca7e64f | Stefan Karpinski | 18 December 2017, 19:01:28 UTC | Merge pull request #16906 from JuliaLang/rf/randjump-genpoly MersenneTwister: implement generation of jump-ahead polynomials | 18 December 2017, 19:01:28 UTC |
7d92d3d | Stefan Karpinski | 18 December 2017, 18:58:03 UTC | Merge pull request #25034 from JuliaLang/jn/require_ji_better2 loading: refactor .ji loading, to handle errors better | 18 December 2017, 18:58:03 UTC |
dcddbca | Jameson Nash | 18 December 2017, 17:55:40 UTC | show: improve handling of . syntax printing (#24994) | 18 December 2017, 17:55:40 UTC |
8b01dfa | Jameson Nash | 18 December 2017, 16:49:12 UTC | make IndexValue act like an AbstractDict (#25114) Since this type acts like an AbstractDict container under iteration, it should also be reasonable to make it also act like an AbstractDict under other common operations (such as calling `pairs` and `getindex`) | 18 December 2017, 16:49:12 UTC |
88cdf44 | Steven G. Johnson | 18 December 2017, 15:39:59 UTC | add :color attribute to IOContext, prefer get(io,:color,false) to Base.have_color (#25067) | 18 December 2017, 15:39:59 UTC |
801dfc4 | Rafael Fourquet | 12 June 2016, 16:13:24 UTC | MersenneTwister: implement generation of jump-ahead polynomials | 18 December 2017, 14:09:20 UTC |
5e50550 | Valentin Churavy | 18 December 2017, 13:28:29 UTC | Merge pull request #25155 from JuliaLang/vc/distributed_topo rename topology master_slave to master_worker | 18 December 2017, 13:28:29 UTC |
16bdf88 | Chris Foster | 18 December 2017, 03:53:50 UTC | Remove from_c caller frames from depwarn attribution Fixes #25130 | 18 December 2017, 10:48:10 UTC |
1b01101 | Chris Foster | 18 December 2017, 10:47:09 UTC | Fix global_logger(new) to return old logger This seems generally more useful than returning the new one. | 18 December 2017, 10:48:10 UTC |
7fb5d27 | Fredrik Ekre | 18 December 2017, 08:00:20 UTC | remove the last eye method (#25153) | 18 December 2017, 08:00:20 UTC |
3cae287 | Stefan Karpinski | 18 December 2017, 07:21:27 UTC | Merge pull request #25102 from JuliaLang/sk/home rename JULIA_HOME => Sys.BINDIR | 18 December 2017, 07:21:27 UTC |
2b23e37 | Sacha Verweij | 18 December 2017, 06:34:05 UTC | Merge pull request #25145 from JuliaLang/ksh/arrdocagain Various doctests for array stuff [ci skip] | 18 December 2017, 06:34:05 UTC |
8ec9274 | Valentin Churavy | 18 December 2017, 04:46:27 UTC | rename topology master_slave to master_worker | 18 December 2017, 05:25:08 UTC |
e83ce4b | Stefan Karpinski | 15 December 2017, 09:42:52 UTC | rename JULIA_HOME => Sys.BINDIR [fix #20899] ENV["JULIA_HOME"] => ENV["JULIA_BINDIR"] | 18 December 2017, 05:09:52 UTC |
4b625bb | kshyatt | 17 December 2017, 21:30:53 UTC | Various doctests for array stuff [ci skip] | 18 December 2017, 05:07:10 UTC |
43eee7f | Stefan Karpinski | 15 December 2017, 23:29:22 UTC | capitalize Sys.CPU_NAME This was the only value exported from Sys with a lowercase name – all other lowercase exports are functions and values are uppercase. | 18 December 2017, 04:24:48 UTC |
d48b9a4 | Stefan Karpinski | 16 December 2017, 00:26:46 UTC | @deprecate_binding: print the new binding by default | 18 December 2017, 04:20:05 UTC |
4d3d49d | Jameson Nash | 18 December 2017, 03:20:35 UTC | implement getfield overloading (#24960) New functions `Base.getproperty` and `Base.setproperty!` can be overloaded to change the behavior of `x.p` and `x.p = v`, respectively. This forces inference constant propagation through get/setproperty, since it is very likely this method will yield better information after specializing on the field name (even if `convert` is too big to make us want to inline the generic version and trigger the heuristic normally). closes #16195 (and thus also closes #16226) fix #1974 | 18 December 2017, 03:20:35 UTC |
a28f495 | Valentin Churavy | 18 December 2017, 03:15:09 UTC | Merge pull request #25126 from JuliaLang/amitm/listenport Distributed: worker to supply a port_hint in the known range | 18 December 2017, 03:15:09 UTC |
d8929e9 | Jameson Nash | 18 December 2017, 02:42:02 UTC | build: enhance usage of libwhich path resolution to respect SONAME (#24916) If the SONAME is available, it directly names the file we want to link to, so use that instead. | 18 December 2017, 02:42:02 UTC |
4817d99 | Fredrik Ekre | 18 December 2017, 01:19:39 UTC | make adjoint! return first argument (#25147) | 18 December 2017, 01:19:39 UTC |
e58b7c7 | Jameson Nash | 12 December 2017, 00:14:28 UTC | loading: refactor .ji loading, to handle errors better | 18 December 2017, 00:58:08 UTC |
943e54a | Sacha Verweij | 17 December 2017, 23:56:59 UTC | Fix location of 0.7 news items accidentally placed in 0.6 section. [ci skip] | 17 December 2017, 23:57:42 UTC |
cfd8d9f | Sacha Verweij | 17 December 2017, 23:36:43 UTC | Move 0.6 NEWS.md section to HISTORY.md. [ci skip] | 17 December 2017, 23:57:35 UTC |
77ef406 | Katharine Hyatt | 17 December 2017, 23:33:50 UTC | Xrefs and a little manual addition about atomic ops (#25120) | 17 December 2017, 23:33:50 UTC |
36b33d7 | Sacha Verweij | 16 December 2017, 02:22:45 UTC | Deprecate .'. | 17 December 2017, 19:13:43 UTC |
24292dd | Sacha Verweij | 16 December 2017, 02:22:32 UTC | Remove special lowering of .' in *, /, and \ expressions to A[t]_(mul|ldiv|rdiv)_B[t]. | 17 December 2017, 19:13:43 UTC |
ae6056a | Sacha Verweij | 16 December 2017, 00:23:15 UTC | Strip .' from test/. | 17 December 2017, 19:13:43 UTC |
3efa9a9 | Sacha Verweij | 15 December 2017, 23:01:44 UTC | Strip .' from stdlib/. | 17 December 2017, 19:09:46 UTC |
fc17e3b | Sacha Verweij | 15 December 2017, 22:31:16 UTC | Strip .' from base/. | 17 December 2017, 19:08:41 UTC |
292a3b4 | Valentin Churavy | 17 December 2017, 18:26:22 UTC | Merge pull request #25142 from JuliaLang/anj/lapack379 Fix occasional test error in lapack.jl | 17 December 2017, 18:26:22 UTC |
a2c97c8 | Rafael Fourquet | 17 December 2017, 17:35:51 UTC | reset GLOBAL_RNG state in/out at-testset for reproducibility (#24445) This is a follow-up on the recently introduced `guardsrand` functionality, first suggested at https://github.com/JuliaLang/julia/pull/16940#issuecomment-226152878. Each `testset` block is now implicitly wrapped in a `guardsrand` block, which is more user-friendly and more systematic (only few users know about `guardsrand`). These are essentially two new features: 1) "in": in base, tests are run with the global RNG randomly seeded, but the seed is printed in case of failure to allow reproducing the failure; but even if the failure occurs after many tests, when they alter the global RNG state, one has to re-run the whole file, which can be time-consuming; with this change, at the beginning of each `testset`, the global RNG is re-seeded with its own seed: this allows to re-run only the failing `testset`, which can be done easily in the REPL (the seeding occurs for each loop in a `testset for`; this also allows to re-arrange `testset`s in arbitrary order w.r.t. the global RNG; 2) "out": a `testset` leaves no tracks of its use of `srand` or `rand` (this "feature" should be less and less needed/useful with the generalization of the use of `testset` blocks). Example: ``` @testset begin srand(123) rand() @testset for T in (Int, Float64) # here is an implicit srand(123), by 1) rand() end rand() # this value will not be affected if the sub-`testset` block # above is removed, or if another rand() call is added therein, by 2) end ``` Note that guardsrand can't be used directly, as then the testset's body is wrapped in a function, which causes problem with overwriting loop variable ("outer"), using `using`, defining new methods, etc. So we need to duplicate guardsrand's logic. | 17 December 2017, 17:35:51 UTC |
a5c9e88 | Rafael Fourquet | 17 December 2017, 09:45:40 UTC | rename some copy! methods to copyto! (#24808) When the deprecation gets lifted, they will be re-enabled with a new meaning. | 17 December 2017, 09:45:40 UTC |
4516fc9 | Andreas Noack | 17 December 2017, 06:51:37 UTC | Make testset readable | 17 December 2017, 06:51:37 UTC |
2195ef9 | Andreas Noack | 17 December 2017, 06:51:11 UTC | Make test tolerance proportional to condition number in a solve test for the Bunch-Kaufman factorization | 17 December 2017, 06:51:11 UTC |
e89a2b8 | Keno Fischer | 16 December 2017, 22:02:53 UTC | Fix missing gc root in compile-all (#25128) ``` /home/keno/julia/src/precompile.c:191:5: note: GC frame changed here JL_GC_PUSH2(&p, &methsig); ^~~~~~~~~~~~~~~~~~~~~~~~~ /home/keno/julia/src/precompile.c:193:5: note: Loop condition is true. Entering loop body while (!incr) { ^ /home/keno/julia/src/precompile.c:194:24: note: Started tracking value here jl_svec_t *p = jl_alloc_svec_uninit(l); ^~~~~~~~~~~~~~~~~~~~~~~ /home/keno/julia/src/precompile.c:195:9: note: Loop condition is true. Entering loop body for (i = 0, idx_ctr = 0, incr = 1; i < l; i++) { ^ /home/keno/julia/src/precompile.c:197:13: note: Taking false branch if (jl_is_uniontype(ty)) { ^ /home/keno/julia/src/precompile.c:195:9: note: Loop condition is false. Execution continues on line 217 for (i = 0, idx_ctr = 0, incr = 1; i < l; i++) { ^ /home/keno/julia/src/precompile.c:217:32: note: Passing non-rooted value as argument to function that may GC methsig = (jl_value_t*)jl_apply_tuple_type(p); ^ ~ ``` Essentially the same bug as #25123 | 16 December 2017, 22:02:53 UTC |