[release-0.5] Backports for 0.5.0-rc4 #18276

tkelman · 2016-08-29T10:53:31Z

will probably be the last RC

@nanosoldier runbenchmarks(ALL, vs = ":release-0.5")

Complex int is GNU extension that Intel compiler does not support currently. #17145 (cherry picked from commit b94da94)

(cherry picked from commit d14155b)

(cherry picked from commit c50e845) ref #18177

…non-Pair return type (cherry picked from commit eb95919) ref #17968

this works to avoid having `Expr(:invoke)` creating unintentional copies of LambdaInfo objects when they show up in the system image fix #18184 (cherry picked from commit b0e692a) ref #18191

…_list many types are easy to compute during deserialization, and don't need the full complexity of the flagref list to handle them some of these are easy to detect ahead-of-time, saving a small amount of deserialization effort (cherry picked from commit 649ce88) ref #18191

fix #15048 (cherry picked from commit 95428ba) ref #18191

Fixes a crash while loading ASTInterpreter (cherry picked from commit 2a8e892) ref #18191

* improve string docs to define String etc * tweak (cherry picked from commit 3ab4d76)

Fixes #18213. (cherry picked from commit 1c5f4ca) ref #18214

fix #18222 (cherry picked from commit 5d2ac0d) ref #18225

fix #18069 (cherry picked from commit 94ce57d) ref #18150

this gives the caller more control over when an how errors get printed (cherry picked from commit 5abccb6) ref #18150

…pile caches this ensures that we only recompile a dependency if either: it can be loaded into the current session, or the user explicitly requests it (via reload) thereby reducing the occurrences and improving the accuracy of the "uuid didn't match" error message :) (cherry picked from commit 0df6e61) ref #18150

fix #18115 (cherry picked from commit 5276a4d) ref #18150

(cherry picked from commit f3fe910)

…#17873) This makes it possible to solve problems with semidefinite B via explicit shift. (cherry picked from commit f95b8b1)

(cherry picked from commit f58aaef)

The permissions specified in MapViewOfFile specify the maximum permissions the file can ever have, which means that even permissions bypass by the debugger is not allowed to write it. Instead, set the maximum permissions to RWX and VirtualProtect down to the permissions we need. This is the same behavior as we use on Mac/Linux. (cherry picked from commit 4f1ba0c) ref #18083

ref staticfloat/homebrew-julia@f57dcc2#diff-633fd9f8448015391714f35e04d39b5bL154 [av skip] (cherry picked from commit ad73afa)

* Preserve meta node and line number info during dead code elimination * Insert `:push_loc` and `:pop_loc` in pairs during lowering Fix #16578 (cherry picked from commit b929c56) ref #18180

Also clean up `julia_const_to_llvm` to avoid boxing due to `getfield`. Fix #18236 (cherry picked from commit c4093ba) ref #18238

Closes #18231. (cherry picked from commit 2961c4d)

(cherry picked from commit 8a70e13) ref #18269

(cherry picked from commit 9743f8d) ref #18269

…omparison (#18234) See #18213. (cherry picked from commit 77176cd)

ufechner7 · 2016-09-05T10:32:11Z

Is it known, why the Travis CI build failed?

davidanthoff · 2016-09-05T20:47:40Z

Is there some way that others can help to get RC4 out? Also, might it make sense to just release RC4 as is, and if more problems show up, just do a RC5? We've been going with RC3 for two weeks now, and essentially that means that a lot of potential testers are not testing the latest bits.

tkelman · 2016-09-05T23:18:20Z

Help determine what's causing the regressions identified by packageevaluator. I'm not tagging an RC with known regressions, so will revert the precompilation changes if that's the necessary fix.

bicycle1885 · 2016-09-06T00:08:04Z

The problem of Cairo.jl seems to have nothing to do with this change. In a clean environment on my OS X of this branch, I saw the same error as JuliaGraphics/Cairo.jl#148. I believe this is a problem of BinDeps.jl. The problem happens on v0.5-RC3, too.

bicycle1885 · 2016-09-06T00:24:11Z

I couldn't reproduce the test failure of Complementarity.jl. It seems to be under active development and there are two minor releases after the Package evaluator (https://github.com/chkwon/Complementarity.jl/releases). So, I guess the problem is not in Julia but in the tests of Complementarity.jl.

lobingera · 2016-09-06T09:19:45Z

@bicycle1885 Could you clarify which Cairo failure you see here? If i look into this #18276 (comment) i get the impression that runtest.jl fails with a signal(11) - and not the BinDeps problem you mention. I had a similar failure trying to build some PR for Cairo and indentified this as a memory management / GC problem, which is mostlikely a Cairo internal problem and was hidden by some 'other' gc behaviour.

(cherry picked from commit 83b0589) ref #18301 Update OpenBLAS to v0.2.19 for Power8. (cherry picked from commit 857469d)

* Avoid aliasing in in UniformScaling*AbstractMatrix ...and remove unnecessary UniformScaling*SparseMatrixCSC methods * Broaden the tests for non-commutative multiplication * Add Quaternion test case for q*[q] and clean up the Quaternion test type (cherry picked from commit cd94c99)

* Fix #18189 (Add deprecation of etc. to NEWS) * Remove mentioning of @static, add as a note * Added reference to Handling Operating System Variation section in manual * Changed reference numbers * Change PR number (cherry picked from commit 1af52d0)

(partial backport of b7c25d1) ref #18334

(cherry picked from commit 00cb3f7) ref #18345

* Removed confusing array of types in readcsv * Restored brackets to `T` in readcsv doc and added clarifying text * genstdlib rst output for docstring change. (cherry picked from commit 22cc7a5)

errors when running test/math.jl and test/misc.jl on the same worker (cherry picked from commit b3d2f30) ref #18373

tkelman · 2016-09-06T18:45:11Z

Am hoping that #18366 will fix the issue. Once that PR has addressed its review I'll backport it in here and run pkgeval again, fingers crossed.

(cherry picked from commit f50dd6a)

tkelman · 2016-09-06T20:13:42Z

@ufechner7

Is it known, why the Travis CI build failed?

I can't recall having seen this error before, but it looks intermittent. Not sure what it is or whether it also would occur on master.

    From worker 9:       * spawn                       [stdio passthrough ok]
    From worker 8:       * sorting               in  85.87 seconds, maxrss  625.10 MB
    From worker 9:  Test Failed
    From worker 9:    Expression: desc == "Pipe(open => active, 0 bytes waiting)"
    From worker 9:     Evaluated: "Pipe(open => active, 9 bytes waiting)" == "Pipe(open => active, 0 bytes waiting)"

vtjnash · 2016-09-06T20:19:41Z

looks like there's a race in that test

tkelman · 2016-09-07T17:32:17Z

Yes. If we see it again, should open an issue for it.

My pkgeval run last night froze on several packages doing Pkg.build on Cairo, but it happened for both rc3 and rc4 so it's not a regression, just a nasty bug that needs fixing. Running again, hopefully won't freeze this time.

also add plenty of helpful debugging information enabled by the environment variable: JULIA_DEBUG_LOADING this bug would cause compilation to become confused and likely tardy at recompiling files when it should that would up with longer dependency chains where one of the files had been updated this is a fix for #18150 (cherry picked from commit 718bd3e)

tkelman · 2016-09-08T02:32:30Z

A fair amount going on here, but does look like the precompilation issue is mostly gone - https://gist.github.com/6715eba41378cb57642b79255d86a723

AmplNLWriter has a bunch of huge C++ test dependencies and most likely ran out of time and got terminated
Unexpected HTTP status code: 500 network failures during clone for Anasol, ApproxFun, GeneralizedSampling, GeneticAlgorithms, and OpenGL (Anasol was the only one of these that wasn't already failing on rc3)
BlsData is flaky and gets a Request to BLS failed with message 'REQUEST_NOT_PROCESSED' failure intermittently, not a Julia problem
Cairo failed with Error in /home/vagrant/julia/bin/julia': double free or corruption (!prev): 0x00000000028464c0 ***\n\nsignal (6): Aborted\nwhile loading /home/vagrant/.julia/v0.5/Cairo/samples/sample_imagepattern.jl, in expression starting on line 29` which is most likely a problem in the package itself, should be looked at but I don't think it's a Julia regression
ImageQuilting failed with a real error that I can reproduce outside of pkgeval TypeError: Tuple: in non-final parameter, expected Type{T}, got Type{Vararg{ParserCombinator.Transform,N}}\n in _include_from_serialized(::String) at ./loading.jl:150\n in _require_from_serialized(::Int64, ::Symbol, ::String, ::Bool) at ./loading.jl:187\n in _require_search_from_serialized(::Int64, ::Symbol, ::String, ::Bool) at ./loading.jl:217 but I think it's the only one
NEOS is flaky and gets a LoadError: XML-RPC failed\n in _send(::NEOS.NEOSServer, ::String) at /home/vagrant/.julia/v0.5/NEOS/src/NEOSServer.jl:46 intermittently, not a Julia problem

lobingera · 2016-09-08T05:07:24Z

@tkelman I agree with your decision that the orignal problem is in the memory management of Cairo.jl (and libcairo). My assumption (without any background on details) is, work on GC has corrected something and now a hidden Cairo.jl problem surfaces. I can not currently spend time on debugging this.

tkelman · 2016-09-08T06:51:53Z

@staticfloat took the Linux buildbots down to rebuild gcc for dmlc/MXNet.jl#138. Not the best timing, that should've waited until after this RC. Can't build binaries until they're back up.

tkelman · 2016-09-09T07:44:06Z

buildbots came back and the post-gcc-upgrade binaries look okay (thanks @staticfloat), so rc4 has been tagged and uploaded

Ehsan Totoni and others added 26 commits August 29, 2016 02:41

ccalltest.c: ifndef Intel compiler for complex int (#18193)

de2e847

Complex int is GNU extension that Intel compiler does not support currently. #17145 (cherry picked from commit b94da94)

Describe negative bit shift counts in NEWS.md (#18188)

ed08c30

(cherry picked from commit d14155b)

ASAN: disable leak detection for flisp.

78096f4

(cherry picked from commit c50e845) ref #18177

Provide a more useful error message for calling map on a Dict with a …

e8285b9

…non-Pair return type (cherry picked from commit eb95919) ref #17968

incremental deserialize: handle LambdaInfo identity uniquing

9bb8af8

this works to avoid having `Expr(:invoke)` creating unintentional copies of LambdaInfo objects when they show up in the system image fix #18184 (cherry picked from commit b0e692a) ref #18191

precompile: major LLVM speed improvement

8884d9d

fix #15048 (cherry picked from commit 95428ba) ref #18191

typemap: handle vararg tuple subtyping in jl_typemap_assoc_by_type

fcd8f2f

Fixes a crash while loading ASTInterpreter (cherry picked from commit 2a8e892) ref #18191

improve string docs to define String etc (#18216)

2634acb

* improve string docs to define String etc * tweak (cherry picked from commit 3ab4d76)

Move specialized isequal definitions to avoid hitting #265

d20ef05

Fixes #18213. (cherry picked from commit 1c5f4ca) ref #18214

inliner needs to do method match tests early

ead4690

fix #18222 (cherry picked from commit 5d2ac0d) ref #18225

fix sort / fallback order for incremental precompile

4568dd1

fix #18069 (cherry picked from commit 94ce57d) ref #18150

change incremental serialize to use deferred errors

a51c614

this gives the caller more control over when an how errors get printed (cherry picked from commit 5abccb6) ref #18150

document __init__ behavior during precompile

746cc0c

fix #18115 (cherry picked from commit 5276a4d) ref #18150

Fix chol for Symmetric(A, :L). (#18142)

c0da8be

(cherry picked from commit f3fe910)

Don't throw in eigs on semidefinite Bs for generalized eigenproblems. (…

aa74d6d

…#17873) This makes it possible to solve problems with semidefinite B via explicit shift. (cherry picked from commit f95b8b1)

add missing "Threads." in example (#18219)

25979af

(cherry picked from commit f58aaef)

hopefully unbreak osx travis llvm linking (#18241)

f35bc7b

ref staticfloat/homebrew-julia@f57dcc2#diff-633fd9f8448015391714f35e04d39b5bL154 [av skip] (cherry picked from commit ad73afa)

Make sure :push_loc meta always has a corresponding :pop_loc

7e2997a

* Preserve meta node and line number info during dead code elimination * Insert `:push_loc` and `:pop_loc` in pairs during lowering Fix #16578 (cherry picked from commit b929c56) ref #18180

Handle VecElement in julia_const_to_llvm

4d8df07

Also clean up `julia_const_to_llvm` to avoid boxing due to `getfield`. Fix #18236 (cherry picked from commit c4093ba) ref #18238

Clarifiy performance results (#18242)

631f752

Closes #18231. (cherry picked from commit 2961c4d)

Speed up indices for SubArray

b9ec3fd

(cherry picked from commit 8a70e13) ref #18269

Use inbounds in reductions with generic iterators

936a9a8

(cherry picked from commit 9743f8d) ref #18269

Test for Dict not triggering a deprecation due to ::Integer==::Char c…

245b9f6

…omparison (#18234) See #18213. (cherry picked from commit 77176cd)

tkelman added this to the 0.5.0 milestone Aug 29, 2016

This was referenced Aug 29, 2016

cache compile dep tracking #18150

Merged

Checklist of steps towards 0.5 release [candidates] #17418

Closed

Fix line number in a doctest

15019c1

ViralBShah mentioned this pull request Sep 3, 2016

Test failure on 0.5rc4 JuliaStats/StatsBase.jl#210

Closed

felipenoris mentioned this pull request Sep 5, 2016

@test is extremely slow in 0.5 for BusinessDays.jl #18361

Closed

ViralBShah and others added 7 commits September 6, 2016 10:52

ppc64 build cleanups

0ef4739

(cherry picked from commit 83b0589) ref #18301 Update OpenBLAS to v0.2.19 for Power8. (cherry picked from commit 857469d)

Fix jl_get_specialization signature mismatch

a9e753b

(partial backport of b7c25d1) ref #18334

Fix docs for isdefined

ee92c24

(cherry picked from commit 00cb3f7) ref #18345

Removed confusing array of types in readcsv help (#18308)

794de80

* Removed confusing array of types in readcsv * Restored brackets to `T` in readcsv doc and added clarifying text * genstdlib rst output for docstring change. (cherry picked from commit 22cc7a5)

Rename a few 'err' variables to avoid invalid redefinition

9601961

errors when running test/math.jl and test/misc.jl on the same worker (cherry picked from commit b3d2f30) ref #18373

rm stray : in manual

becc272

(cherry picked from commit f50dd6a)

vtjnash and others added 2 commits September 7, 2016 16:14

fix line numbers in a doctest

67e398b

tkelman mentioned this pull request Sep 8, 2016

OSX julia-0.5 segfault on sample_imagepattern JuliaGraphics/Cairo.jl#146

Closed

tkelman merged commit e1d2965 into release-0.5 Sep 8, 2016

tkelman deleted the tk/backports-0.5.0-rc4 branch September 8, 2016 05:47

tkelman mentioned this pull request Sep 8, 2016

TypeError precompiling ImageQuilting.jl Tuple: in non-final parameter, expected Type{T}, got Type{Vararg{ParserCombinator.Drop,N}} #18404

Closed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[release-0.5] Backports for 0.5.0-rc4 #18276

[release-0.5] Backports for 0.5.0-rc4 #18276

tkelman commented Aug 29, 2016

ufechner7 commented Sep 5, 2016

davidanthoff commented Sep 5, 2016

tkelman commented Sep 5, 2016

bicycle1885 commented Sep 6, 2016

bicycle1885 commented Sep 6, 2016

lobingera commented Sep 6, 2016 •

edited

Loading

tkelman commented Sep 6, 2016

tkelman commented Sep 6, 2016

vtjnash commented Sep 6, 2016

tkelman commented Sep 7, 2016 •

edited

Loading

tkelman commented Sep 8, 2016

lobingera commented Sep 8, 2016

tkelman commented Sep 8, 2016

tkelman commented Sep 9, 2016

[release-0.5] Backports for 0.5.0-rc4 #18276

[release-0.5] Backports for 0.5.0-rc4 #18276

Conversation

tkelman commented Aug 29, 2016

ufechner7 commented Sep 5, 2016

davidanthoff commented Sep 5, 2016

tkelman commented Sep 5, 2016

bicycle1885 commented Sep 6, 2016

bicycle1885 commented Sep 6, 2016

lobingera commented Sep 6, 2016 • edited Loading

tkelman commented Sep 6, 2016

tkelman commented Sep 6, 2016

vtjnash commented Sep 6, 2016

tkelman commented Sep 7, 2016 • edited Loading

tkelman commented Sep 8, 2016

lobingera commented Sep 8, 2016

tkelman commented Sep 8, 2016

tkelman commented Sep 9, 2016

lobingera commented Sep 6, 2016 •

edited

Loading

tkelman commented Sep 7, 2016 •

edited

Loading