b24213b | Elliot Saba | 23 November 2014, 20:19:28 UTC | Tag v0.3.3 | 23 November 2014, 20:19:28 UTC |
e596df2 | Ivar Nesje | 23 November 2014, 16:55:08 UTC | Merge pull request #9120 from JuliaLang/sf/0.3.3-testing Final tweaks before 0.3.3 tagging | 23 November 2014, 16:55:08 UTC |
02e57db | Elliot Saba | 23 November 2014, 08:43:53 UTC | Remove sys.dll on windows when running `make dist`, until related issues are solved (#8895, among others) | 23 November 2014, 08:43:53 UTC |
18ad3e5 | Elliot Saba | 23 November 2014, 05:39:03 UTC | Change link to Unicode overview to archive.org | 23 November 2014, 08:30:39 UTC |
c18454d | Elliot Saba | 23 November 2014, 05:25:46 UTC | Fix sphinx warning for `chmod` docs. Without this fix, the following warning appears: ``` Warning, treated as error: /Users/vagrant/src/julia-0.3/doc/stdlib/base.rst:5388: WARNING: Inline literal start-string without end-string. ``` | 23 November 2014, 08:30:25 UTC |
aafdce2 | tkelman | 26 July 2014, 13:13:43 UTC | Allow deletion of read-only files on Windows | 23 November 2014, 08:30:25 UTC |
9fac21b | tkelman | 13 July 2014, 22:16:28 UTC | Simple implementation of chmod(path, mode) currently only integer modes, can do string modes like +x later fixes #7574 Conflicts: src/jl_uv.c | 23 November 2014, 08:30:25 UTC |
ac11aad | ivarne | 23 November 2014, 08:27:01 UTC | update helpdb.jl | 23 November 2014, 08:27:01 UTC |
10f9289 | Tony Kelman | 22 November 2014, 19:42:41 UTC | Merge pull request #9112 from JuliaLang/tk/fix-pkg-testdir use tempdir instead of tempname in test/pkg.jl | 22 November 2014, 19:42:41 UTC |
dd6cd5a | Tony Kelman | 22 November 2014, 19:17:04 UTC | use tempdir instead of tempname in test/pkg.jl since tempname creates a file on windows, but the goal here is creating a directory. also don't try to delete nonexistent files during cleanup | 22 November 2014, 19:17:04 UTC |
83f8018 | ivarne | 22 November 2014, 16:32:04 UTC | Fix another issue with the backport of #8889 | 22 November 2014, 16:32:04 UTC |
aa5c36d | Ivar Nesje | 22 November 2014, 11:34:19 UTC | Fis broken backport We must be more careful when backporting now that we have differences between 0.3 and 0.4 | 22 November 2014, 11:34:19 UTC |
9804226 | jake bolewski | 20 November 2014, 22:08:06 UTC | check for valid git version number in pkg Conflicts: base/pkg/dir.jl | 21 November 2014, 22:13:26 UTC |
33192cb | Gunnar Farnebäck | 20 November 2014, 22:50:24 UTC | Return type stability fix Missing type specification in zeros calls, causing inconsistent return type when T != Float64. (cherry picked from commit 83494d3a3afff902c88ebf49363661f7377f5f4e) | 21 November 2014, 20:47:58 UTC |
e58d214 | Ivar Nesje | 20 November 2014, 07:13:23 UTC | Merge pull request #9074 from JuliaLang/cjh/gitignore-0.3 Add objconv to deps/.gitignore | 20 November 2014, 07:13:23 UTC |
3c8fccb | Jiahao Chen | 20 November 2014, 02:59:14 UTC | Add objconv to deps/.gitignore | 20 November 2014, 02:59:14 UTC |
75b347b | Milan Bouchet-Valat | 16 November 2014, 16:32:51 UTC | Document return type for middle(x) The special cases give the same result as the general rule, but are faster. Fix the docs and explain the rule as regards the result type. (cherry picked from commit e955abdf3f45c7afe736937235aac5700916821c) Conflicts: base/statistics.jl Improve wording of docs for middle() Cf. #9032. (cherry picked from commit b5f762c9833c48e5d2dc4460d2c8605e99b296ab) | 19 November 2014, 10:23:36 UTC |
67e2e2b | Remi Berson | 02 November 2014, 09:42:39 UTC | Document middle manual backport of ed834102cde9ae9ed3e8b135f2aafafe33f495b8 Add a note on return type in documentation of middle (cherry picked from commit 35b8a22d950c87f20f5b5bd6586100a84bffc627) Improve documentation of middle with suggestion of @pao (cherry picked from commit d9a68959b56e706f3d18ebdce204ed404fad110d) | 19 November 2014, 10:23:27 UTC |
c01e415 | Tony Kelman | 19 November 2014, 09:52:17 UTC | Merge pull request #9050 from JuliaLang/tk/backport-8734 RFC: Should we backport #8734 to release-0.3 ? | 19 November 2014, 09:52:17 UTC |
76522d7 | Mike Nolta | 10 November 2014, 00:17:00 UTC | fix #8954 (BigFloat printf buffer overflow) backport of a015cce7fdc2bfc116530a24bc1adabc419134ab Conflicts: base/printf.jl | 19 November 2014, 04:54:09 UTC |
16146c8 | Erik Schnetter | 16 November 2014, 04:15:45 UTC | Document foldl and foldr implementation backport of 7c8bf1304458004a9e37cb2eba2cf507aa0fcdea Document mapreduce and friends Update documentation for reduce, foldl, and foldr. Add mapreduce documentation for variant with v0. Add mapfoldl, mapfoldr documentation. backport of 2c6707426cf2ccd8985d3c0b516afc452d3e956d | 19 November 2014, 04:48:33 UTC |
1f68fa8 | Steven G. Johnson | 18 November 2014, 19:23:53 UTC | clarify manual on isequal and hash (fix #6833) backport of e2f71663e4615b105b525cf25c90fbdfb371bd0b | 19 November 2014, 04:44:31 UTC |
6433492 | Milan Bouchet-Valat | 11 November 2014, 15:29:39 UTC | Improve docs for tempname() and tempdir() backport of 74f67560f4428b1febddcd45972281efa46923ce | 19 November 2014, 04:42:55 UTC |
4c56a6b | Steven G. Johnson | 18 November 2014, 16:48:01 UTC | some clarifications on Fortran mangling backport of 045d95078cf73b3b199f3f89a0993b7adf571462 | 19 November 2014, 04:37:45 UTC |
b0977da | Isaiah | 15 November 2014, 15:04:03 UTC | Add note on Fortran name-mangling, fixes #9022 backport of a06e9385490266f794556ea9e4c7a3b56235cf65 | 19 November 2014, 04:37:24 UTC |
e93af36 | Mike Nolta | 14 November 2014, 18:27:12 UTC | fix jl_fs_write type sig uv_fs_write:offset is int64_t, not size_t. Closes nolta/Winston.jl#182. backport of c6bdbfebd5eba9889b9138a7294319da0ba2c526 Conflicts: base/fs.jl src/jl_uv.c | 19 November 2014, 04:19:38 UTC |
28e5f11 | Jeff Bezanson | 18 November 2014, 15:30:56 UTC | don't use the word "copy" in docs for `fill`; the point is that there is no copy(). backport of d9fb7e02b8f48a83bf1db33788fb7083dc6b4e6d | 19 November 2014, 03:16:37 UTC |
84f6a45 | Milan Bouchet-Valat | 08 November 2014, 16:18:01 UTC | Use tempname() instead of custom file path in test/pkg.jl Fixes calling Base.runtests("pkg") when Julia is installed in a prefix where typical user does not have write access. backport of 8e75fd2d2921b3908864aff4b714e26e6791d0a6 | 19 November 2014, 03:13:20 UTC |
ceb9632 | Gustavo Goretkin | 12 November 2014, 19:10:39 UTC | Update intfuncs.jl otherwise, even tough `pad` should be non-negative julia> digits(5,2,uint64(8)) ERROR: `digits` has no method matching digits(::Int64, ::Int64, ::Uint64) backport of 07f26b491ee2091afd1b06924899a7514abe03fa | 19 November 2014, 03:10:35 UTC |
ab7c6eb | Viral B. Shah | 11 November 2014, 06:52:39 UTC | Better fix for #8976 that takes care of both, conj and conj! | 18 November 2014, 19:04:21 UTC |
257c430 | Jeff Bezanson | 18 November 2014, 17:15:33 UTC | Merge pull request #9055 from dhoegh/shell_completion_fix_back_port Backport of #8995 to 0.3 | 18 November 2014, 17:15:33 UTC |
3197b59 | Daniel Høegh | 12 November 2014, 18:31:29 UTC | Fix that REPL crashes in shell mode on $a<tab> and added a test. | 18 November 2014, 12:22:12 UTC |
4611fa8 | Tim Holy | 18 November 2014, 12:05:04 UTC | Clean up devdocs indexing | 18 November 2014, 12:05:04 UTC |
658c865 | timholy | 14 November 2014, 11:05:19 UTC | Documentation: add info about argument to `jl_breakpoint` backport of 3340b2f1bebaa20c355f46cbd793842f24c626df | 18 November 2014, 09:23:05 UTC |
507a4c6 | timholy | 14 November 2014, 02:54:45 UTC | Document all sorts of C debugging tips backport of cbb4bc3ffadc49c828b34ff69a3948e00cbb60e6 Conflicts: doc/index.rst | 18 November 2014, 09:17:29 UTC |
8243c5b | Tony Kelman | 22 October 2014, 15:26:33 UTC | add .tar.xz and .zip files to source-dist (backport of 774f3c93f0cec1eaeebce63a8d45808a55e496df) | 18 November 2014, 08:11:44 UTC |
d297ffb | Tony Kelman | 21 October 2014, 18:50:22 UTC | add osx-only objconv dependency for renaming symbols (backport of fef2954881dc2dac47853b3d72cce15cf69374ed) Conflicts: deps/Makefile Hopefully fix #8842, objconv dependency not right (backport of dbcd50328314d2f632f4927466952868f581751d) We needed to move the objconv rules to above the openblas rules Fixes #8842 (backport of 2579a649fe8a6d0ee8b7d925e343527a4c8b622b) Conflicts: deps/Makefile | 18 November 2014, 08:11:44 UTC |
bd9b63a | Tony Kelman | 18 November 2014, 07:45:02 UTC | add blasfunc for appending suffix to all blas and lapack symbols (manual backport of 02ff76249c3f43247b9898db9edee48af73f1229) | 18 November 2014, 08:11:44 UTC |
12797db | Tony Kelman | 19 October 2014, 20:59:55 UTC | add a separate blas_vendor for suffixed openblas (backport of 7f596d460944bd85ddf7379b70f37507ae3eda8c) | 18 November 2014, 07:21:42 UTC |
066825e | Tony Kelman | 19 October 2014, 14:27:59 UTC | Patch openblas to add 64_ suffix when using 64 bit ints objconv cannot do in-place modification of files dont overwrite static library with objcopy on freebsd either fix linking of arpack and suitesparse with 64_ blas suffix the patched xlahqr2 was using UPPERCASE function names, and the probes during arpack configure were using the C compiler, sgemm_ and cheev_ need to enable c preprocessor, and disable syntax error on long lines SPQR also needed -DSUN64 (backport of 1ef22d21e5581ac9f2b80b7de6183f8c77e8a56e) | 18 November 2014, 07:21:26 UTC |
c14aeab | Will Kearney | 13 October 2014, 01:44:56 UTC | Fix bug in issym(A::BitMatrix) numnz -> countnz backport of 7ba14fe8dc25286df86c03c10f5d3d9260558254 Add issym test for bitarrays backport of 85b015f8dfc089698db3904a89a25b14d49552f7 | 18 November 2014, 06:58:27 UTC |
8425d55 | Steven G. Johnson | 10 November 2014, 17:32:11 UTC | fix #8968: abs(big(-0.0)) backport of a42a54bf5ba439850d4d78dc013c397e49cac79b | 18 November 2014, 06:42:14 UTC |
ea4f2c9 | Tony Kelman | 05 November 2014, 02:27:31 UTC | Document Cygwin-to-MinGW cross compilation Replace old-and-busted MSYS1 instructions, no one should be using MSYS1 any more. backport of 616153c7e64b32744f3ca964062e00154b3aa963 add a call to env in the git troubleshooting alias backport of 5576084b3ca764eb9fa1e69554e7fb24d944193e | 18 November 2014, 06:35:45 UTC |
990bcfb | Tony Kelman | 05 November 2014, 01:04:17 UTC | check for and error early with cygwin or msys-target compilers these are incapable of building libuv, but can build everything else backport of ec1cdaf161a5ba2de83070f955552139492e9136 set XC_HOST to `uname -m`-w64-mingw32 by default for cygwin builds backport of 123de5ac05cf8e1669f1080286c0d0229d4774b2 | 18 November 2014, 06:35:39 UTC |
b84b7a6 | Pontus Stenetorp | 10 November 2014, 04:08:58 UTC | Additional colour resets for the ASCII logo backport of 3dba544808f4c2aa3a95ee6899fc11db6ca4b1fc | 18 November 2014, 06:12:09 UTC |
cdf1f97 | Ondrej Svitek | 12 November 2014, 16:22:22 UTC | README.windows: corrected git config args backport of 6af22f7c7642a929a5c36c7d23c9043592463ebe | 18 November 2014, 06:07:56 UTC |
8119a29 | Tony Kelman | 01 November 2014, 15:07:59 UTC | add a test for getindex(::Number, ::Real) ref b718cbc72e90 backport of e5e489d8bbc83ff61d648ad58612b312c92f1c12 | 18 November 2014, 05:55:40 UTC |
2c90930 | Tony Kelman | 07 November 2014, 23:59:01 UTC | suggest running Pkg.update() in error message backport of 3337117257d46a1b8e249e958d9b62abc20e009a | 18 November 2014, 05:46:48 UTC |
be36f9e | Davide Lasagna | 08 November 2014, 14:14:32 UTC | Name atan2 arguments consistently Doing `methods(atan2)` reveals an inconsistency in the names of the arguments x and y. The documentation about atan2 says `atan2(y, x)`, but the names in the functions are inverted. backport of c5d868f1832381d41dd14ad2efe3c706d1b7dc69 Changed argument names for atan2 in float16.jl backport of 0ed59050d6cb2cb489eed004305a8a759cb36cfb | 18 November 2014, 05:30:47 UTC |
76edcf2 | Amit Murthy | 15 November 2014, 07:15:17 UTC | doc: further explain fill/fill! backport of 52598ff0103d089282a6a49c2d5dc6d3f2ebff5c | 18 November 2014, 05:15:41 UTC |
efd5966 | Jameson Nash | 16 November 2014, 17:02:51 UTC | fix #8347 backport of 4547cb70699512 | 18 November 2014, 05:00:47 UTC |
6b76e19 | Elliot Saba | 30 October 2014, 23:33:26 UTC | Setting OPENBLAS_TARGET_ARCH should disable OPENBLAS_DYNAMIC_ARCH by default | 18 November 2014, 03:03:00 UTC |
cb6a5ee | Elliot Saba | 14 November 2014, 23:32:48 UTC | Fix OSX permissions | 17 November 2014, 21:43:10 UTC |
5c089db | Jiahao Chen (陈家豪) | 14 November 2014, 20:25:26 UTC | Backport microbenchmark updates to release-0.3 Ref: #4821 #8847 | 14 November 2014, 20:25:26 UTC |
42084f3 | Zydrunas Gimbutas | 22 August 2014, 23:45:08 UTC | Adjust the default threshold parameters for the least squares The present tolerance is larger than necessary in the implementation of ldiv operator which can lead to loss of accuracy. Set the default tolerance parameter that is compatible with the conservative MATLAB estimate for determining the numerical rank of A, i.e., tol= eps()*maximum(size(A)). In addition, fix the implementation of the SVD based least squares, and increase the default tolerance parameter for the Moore–Penrose pseudoinverse which can not be determined to full machine precision for ill-conditioned matrices. | 13 November 2014, 15:00:47 UTC |
9c5d8fb | Elliot Saba | 11 November 2014, 00:51:38 UTC | Merge pull request #8970 from goretkin/release-0.3 log10(float16(100)) doesn't cause infinite recursion anymore (along with other functions) | 11 November 2014, 00:51:38 UTC |
21f2894 | goretkin | 10 November 2014, 23:23:36 UTC | log10(float16(100)) doesn't cause infinite recursion anymore | 10 November 2014, 23:23:36 UTC |
201c916 | timholy | 24 October 2014, 10:28:33 UTC | Raise informative error for exponentiation to negative power (see #3024) (cherry picked from commit 79c9dfc5864df04fce95f5c5b7a328795f3fda85) make #8784 work on Mac (cherry picked from commit d2252d08d9b410d45a6ef955f6839ef2bcbe6ddf) | 07 November 2014, 13:12:07 UTC |
f218d2e | ivarne | 07 November 2014, 12:44:24 UTC | Int(length(A)) must be int(length(A)) for 0.3 | 07 November 2014, 12:44:24 UTC |
245b9f6 | ivarne | 03 November 2014, 19:44:19 UTC | Backport Fix for #8871 Mapreduce didn't work with `Range{BigInt}`, because the length wasn't a `Int`, because of potential overflow. This patch just converts the length to Int with a checked conversion, because a mapreduce on larger arrays will probably never finish anyway. Also fix some broken test of prod (cherry picked from commit 94934658e5a8d520a999b552c7116434ce7cc4c4) Fix test for 32 bit The test added in #8893 shouldn't use typemax(Int), but typemax(Int64) (cherry picked from commit 49083be6e7c4ceaa3f457afd1d126084c18615d7) | 07 November 2014, 11:48:57 UTC |
ba119f9 | pabloferz | 25 October 2014, 18:03:58 UTC | Backport #8813 Create Complex.java Class for Complex numbers and simple operations on them. (cherry picked from commit dd2cec8fd8ab5f094def189e6dd891632d59e9cc) Update PerfBLAS.java Change the mandel method to make use of the Complex type removing the manual inlining and strength reduction of complex number operations. (cherry picked from commit fc3223381bcef77132b693720dca7daa7b424ffb) Update Complex.java (cherry picked from commit ed5a00b3d4a4a1f8f93b63314ea952ac5108073e) Update PerfBLAS.java (cherry picked from commit a6e8946ca9874d68a0a3be83b48a98de018f4683) | 07 November 2014, 11:45:43 UTC |
f9dcf0d | Jeff Bezanson | 30 October 2014, 05:12:06 UTC | fix #8851, type intersection bug in Grid.jl tests exposed by `call` overloading (cherry picked from commit da6087eed85253bbe12b075315e2ee87fbf45ef2) Conflicts: test/core.jl | 07 November 2014, 11:42:08 UTC |
478ac06 | Jeff Bezanson | 24 October 2014, 17:32:28 UTC | fix #8768, make `isimmutable` a generic function (cherry picked from commit 7efbf2005b233ad9f26eda123ce4d4213432a9b0) | 07 November 2014, 10:29:49 UTC |
48facaf | brk00 | 20 October 2014, 03:09:43 UTC | Simplify conditionals on edit() and other minor changes on interactiveutil.jl Ref: #8738 (cherry picked from commit 318e03bd5c0cddb1ecdd6907225e741532bf282a) | 07 November 2014, 10:25:19 UTC |
c3666a1 | Andreas Noack | 03 November 2014, 14:56:43 UTC | Fix #8857, i.e. element type promotion in \(SymTridiagonal) Conflicts: base/linalg/tridiag.jl | 05 November 2014, 18:14:56 UTC |
fe82bd5 | Andreas Noack | 05 November 2014, 15:47:41 UTC | Fix inv(Hermitian) and let it return a Hermitian type. Make fallback \ call factorize and add factorize noops for types that can already be considered factorized. Add dummy A_ldiv_B! to cholmod.jl to make fallback \ work. Conflicts: base/linalg/generic.jl base/linalg/symmetric.jl | 05 November 2014, 17:07:47 UTC |
92be3f5 | Tony Kelman | 03 November 2014, 20:06:55 UTC | Merge pull request #8885 from dhoegh/backport_#8693 Backport of #8693 to 0.3 | 03 November 2014, 20:06:55 UTC |
64d5005 | Andreas Noack | 03 November 2014, 17:53:21 UTC | Define eltype for Factorizations | 03 November 2014, 17:55:23 UTC |
417b50a | Tim Holy | 03 November 2014, 11:32:09 UTC | Merge pull request #8821 from JuliaLang/teh/vecmat Relaxed dimensionality requirements for various LinAlg operations (backport 0.3) | 03 November 2014, 11:32:09 UTC |
9b2f2be | Andreas Noack | 02 November 2014, 19:18:33 UTC | In \, only estimate rank and use complete orthogonal factorization for pivoted qr factorizations. | 02 November 2014, 19:27:04 UTC |
ed481ea | Jeff Bezanson | 28 October 2014, 17:06:51 UTC | need to initialize new gc root in jl_compute_type_union (cherry picked from commit 3834eeb820a355b307a78d8c25573a1e7baf91ee) Ref #8819 | 01 November 2014, 15:20:43 UTC |
da878b3 | Oscar Blumberg | 26 October 2014, 20:04:07 UTC | Root a forgotten tuple (cherry picked from commit 0840a0e2a177e3a9f793535988d8295d89f1af4f) (cherry picked from commit 75cff0337de1aa287671cf72633e92b0a5f63035) (cherry picked from commit 36e11eda1aac918f76711e65e16ae70b08e9ce55) Backport of #8819 | 01 November 2014, 14:32:54 UTC |
d8b38aa | Jeff Bezanson | 31 October 2014, 22:43:37 UTC | fix #8865, comprehension evaluated in wrong module this was because the REPL task was still "in" the old Main module, and `expand` depends on the current module but doesn't take a module argument. fortunately there doesn't seem to be any need to call `expand` explicitly here. (cherry picked from commit 6b2e83fe2d133469a686bb1bb87aca8785c42671) | 01 November 2014, 12:17:39 UTC |
003d9da | Daniel Høegh | 18 October 2014, 20:25:12 UTC | Fixes #8483 by correcting the auto completion on windows added some test cases + fixed an error that were discovered when doing `using B<tab>` before Pkg.init(). | 29 October 2014, 14:12:37 UTC |
bdd4597 | StephenVavasis | 12 October 2014, 03:02:55 UTC | Update mathematical-operations.rst I added a note that the updating-operators can change the type. This is implicit in the current manual, but it nonetheless threw me for a loop the first time it happened to me (this is different, e.g., from C++), so I thought it should be made explicit. (cherry picked from commit e244107eb024444c2e29f52953f7e630a62f1942) Simplify wording and rework example as doctest (cherry picked from commit 7fdc860b6347461f929f81482715dc747e46865a) Ref: #8662 | 28 October 2014, 09:26:38 UTC |
3452c5f | Viral B. Shah | 28 October 2014, 02:00:09 UTC | Merge branch 'release-0.3' of github.com:JuliaLang/julia into release-0.3 | 28 October 2014, 02:00:09 UTC |
c7ba781 | Viral B. Shah | 28 October 2014, 01:59:42 UTC | Add checksums for openblas 0.2.12 and remove old ones. | 28 October 2014, 01:59:42 UTC |
3ad21b0 | jake bolewski | 26 October 2014, 02:44:39 UTC | fix bug for getindex(::Number, ::Real) (cherry picked from commit b718cbc72e90e28f8df514e1a320fc2eb4f9279c) | 27 October 2014, 20:17:04 UTC |
3816e88 | Simon Byrne | 21 October 2014, 14:21:11 UTC | Improve speed of `ndigits0z(::BigInt)` using logarithms (#8743). (cherry picked from commit 5445e05e5ef47835d13496c2379b7edbdcf73cd6) Conflicts: base/gmp.jl From #8752 | 27 October 2014, 20:15:59 UTC |
76a8cc1 | Andreas Noack | 08 October 2014, 18:08:12 UTC | Let linalg error messages start with small letters | 27 October 2014, 16:04:16 UTC |
b7b3e28 | Viral B. Shah | 27 October 2014, 14:57:50 UTC | Bump OpenBLAS to v0.2.12, which brings significant performance updates on newer architectures. | 27 October 2014, 14:57:50 UTC |
e0224de | timholy | 27 October 2014, 12:17:24 UTC | More performant solution for transpose! with Vectors (see #8726) | 27 October 2014, 12:28:50 UTC |
a3b380b | timholy | 18 October 2014, 18:25:48 UTC | Relax dimensionality requirements in `A*_mul_B*!` and `transpose!` Conflicts: base/linalg/matmul.jl | 27 October 2014, 12:27:56 UTC |
e92b32c | Viral B. Shah | 25 October 2014, 16:56:15 UTC | Fix #8777 | 25 October 2014, 16:57:01 UTC |
52e33ab | Arch D. Robison | 24 October 2014, 14:54:21 UTC | Merge pull request #8804 from ArchRobison/adr/goto Warn that break, goto, or continue in an @simd loop may cause wrong results | 24 October 2014, 14:54:21 UTC |
78ea467 | Arch D. Robison | 16 October 2014, 19:46:11 UTC | Warn that break, goto, or continue in an @simd loop may cause wrong results. (cherry picked and edited from commit a3a0d6c6d794fd52c29901ed5699162bf887d482) ref: #8704 | 24 October 2014, 14:45:07 UTC |
142798d | timholy | 26 August 2014, 14:38:23 UTC | Performance tips: describe atsign-time and importance of allocation (cherry picked from commit 38f637e273425bc3f32657ed6c08bb218191f897) ref: #8145 | 23 October 2014, 10:59:37 UTC |
5dda348 | Tony Kelman | 15 October 2014, 06:45:13 UTC | fix #8678, reading large files on windows read() on windows takes a uint, so apply a similar io limit per call as done on mac (directly reusing the same code would give a compiler warning, since long is 32 bits on windows) (cherry picked from commit 46e15987e0fb0c3d8f20b791252ab6846df90ceb) ref #8689 | 22 October 2014, 19:35:18 UTC |
b1fc473 | Stefan Karpinski | 20 October 2014, 17:52:59 UTC | ndigits0z(n::BigInt, b): avoid expensive off-by-one check for b=2^k. Fixes #8727, but it's kind of a bandaid since for non-power-of-two bases ndigits0z is still slow due to GMP's inexcusable laziness. At least for power-of-two bases, however, BigInt hashing is now about 3x faster and allocates no memory. (cherry picked from commit 19eb2faf3c80dd327f3c711f614d7610674a2d97) | 22 October 2014, 19:03:42 UTC |
62d83b8 | Ivar Nesje | 22 October 2014, 09:01:29 UTC | Merge pull request #8766 from ivarne/no_fork [release-0.3] Stop thinking about release-0.3 as a fork | 22 October 2014, 09:01:29 UTC |
4cb1e20 | ivarne | 22 October 2014, 08:42:37 UTC | Stop thinking about release-0.3 as a fork This changes the splach screen to the more reasonable view that treats release-0.3 as the master branch. _ _ _ _(_)_ | A fresh approach to technical computing (_) | (_) (_) | Documentation: http://docs.julialang.org _ _ _| |_ __ _ | Type "help()" for help. | | | | | | |/ _` | | | | |_| | | | (_| | | Version 0.3.2-pre+2 (2014-10-22 01:21 UTC) _/ |\__'_|_|_|\__'_| | Commit 6babc84 (0 days old release-0.3) |__/ | x86_64-apple-darwin13.4.0 | 22 October 2014, 08:44:22 UTC |
17aa944 | ivarne | 22 October 2014, 08:40:00 UTC | Begin the march towards 0.3.3 with a "-pre" mark | 22 October 2014, 08:40:00 UTC |
6babc84 | Simon Kornblith | 12 May 2014, 02:27:05 UTC | Fix #6372 This adds the package to REQUIRE on Pkg.clone() if it exists in METADATA. This means that calling Pkg.clone() on a repo that's on an old tag will result in the package getting updated to the latest release, but that seemed preferable to making Pkg.clone() on a package that's on a tag result in a pinned package. | 22 October 2014, 01:21:11 UTC |
1ac3b12 | Simon Kornblith | 22 October 2014, 01:18:53 UTC | Update docs for checknan keyword Ref #8598, cc @ivarne | 22 October 2014, 01:18:53 UTC |
8227746 | Elliot Saba | 21 October 2014, 20:18:59 UTC | Tag v0.3.2 | 21 October 2014, 22:05:51 UTC |
7b1eb64 | Elliot Saba | 21 October 2014, 20:21:42 UTC | Fix typo in get-arpack | 21 October 2014, 22:05:41 UTC |
55f34d7 | Steven G. Johnson | 20 October 2014, 15:21:34 UTC | fix #8698: cp(src,dst) should truncate dst if it already exists | 21 October 2014, 20:14:56 UTC |
8383227 | Milan Bouchet-Valat | 21 October 2014, 07:50:41 UTC | Merge pull request #8748 from dhoegh/backport_of_#8725 Backport of #8725 Fix so `@edit` works on windows with the default editor. | 21 October 2014, 07:50:41 UTC |
c8695e2 | Daniel Høegh | 21 October 2014, 06:45:11 UTC | Backport of #8725 Fix so `@edit` works on windows with the default editor + line number is printed to the user when it is not passed to the editor. | 21 October 2014, 06:45:11 UTC |
e1047d8 | Jeff Bezanson | 19 October 2014, 19:49:43 UTC | fix a missing GC root in codegen | 20 October 2014, 05:05:47 UTC |