Loosen searchsorted* index type (fixes #30763) #31633

kmsquire · 2019-04-06T08:49:12Z

Might also address searchsortedfirst fails with non-Int indices #31618
Types of start and stop indicies are still restricted to
Union{Int32,Int64} and must be the same

kmsquire · 2019-04-08T17:32:55Z

There were some suggestions left by @StefanKarpinski in 60fc1f9 that were lost because of my force push. Will address those, and loosen the types further to accommodate #31618.

kmsquire · 2019-04-08T17:38:11Z

@StefanKarpinski, one of your recommendation was to change T(1) to one(T). I actually had that at first, but it failed during build with an error that one was not available (i.e., this file is compiled before one is available).

I'm inclined to leave it as T(1), but let me know if you have a different suggestion.

kmsquire · 2019-04-08T19:10:19Z

Tried using one again, just to make sure. The error is here:

error during bootstrap:
LoadError(at "compiler/compiler.jl" line 3: LoadError(at "compiler/bootstrap.jl" line 8: UndefVarError(var=:one)))

I'll change it back to T(1)

StefanKarpinski

Looks good to me. Can be merged whenever CI passes.

kmsquire · 2019-04-08T22:05:52Z

Travis and AppVeyor pass.

Buildbot tests seem unrelated (Cc: @staticfloat):

buildbot/testter_macos64:

Error in testset InteractiveUtils:
Error During Test at /Users/julia/buildbot/worker-tabularasa/tester_macos64/build/share/julia/test/testdefs.jl:20
  Got exception outside of a @test
  Error while loading expression starting at /Users/julia/buildbot/worker-tabularasa/tester_macos64/build/share/julia/stdlib/v1.2/InteractiveUtils/test/runtests.jl:386
  caused by [exception 1]
  failed process: Process(`pbcopy`, ProcessExited(1)) [1]

buildbot/tester_win32:

Error in testset FileWatching:
Error During Test at C:\cygwin\home\Administrator\buildbot\worker-tabularasa\tester_win32\build\share\julia\stdlib\v1.2\FileWatching\test\runtests.jl:238
  Test threw exception
  Expression: ispath(tr[1]) && ispath(tr[2])
  MethodError: no method matching joinpath(::EOFError)
  Closest candidates are:
    joinpath(!Matched::String, !Matched::String) at path.jl:266
    joinpath(!Matched::AbstractString) at path.jl:248
    joinpath(!Matched::AbstractString, !Matched::AbstractString) at path.jl:275

buildbot/tester_win64:

Error in testset spawn:
Test Failed at C:\cygwin\home\Administrator\buildbot\worker-tabularasa\tester_win64\build\share\julia\test\spawn.jl:366
  Expression: desc == "Pipe($(infd) open => $(outfd) active, 0 bytes waiting)"
   Evaluated: "Pipe(Base.Libc.WindowsRawSocket(0x0000000000000404) open => Base.Libc.WindowsRawSocket(0x00000000000003fc) active, 9 bytes waiting)" == "Pipe(Base.Libc.WindowsRawSocket(0x0000000000000404) open => Base.Libc.WindowsRawSocket(0x00000000000003fc) active, 0 bytes waiting)"
Error in testset FileWatching:
Test Failed at C:\cygwin\home\Administrator\buildbot\worker-tabularasa\tester_win64\build\share\julia\stdlib\v1.2\FileWatching\test\runtests.jl:179
  Expression: #= C:\cygwin\home\Administrator\buildbot\worker-tabularasa\tester_win64\build\share\julia\stdlib\v1.2\FileWatching\test\runtests.jl:179 =# @elapsed(#= C:\cygwin\home\Administrator\buildbot\worker-tabularasa\tester_win64\build\share\julia\stdlib\v1.2\FileWatching\test\runtests.jl:179 =# @test(watch_folder(dir, 0) == ("" => FileWatching.FileEvent()))) <= 2
   Evaluated: 61.374476086 <= 2
Error in testset FileWatching:
Test Failed at C:\cygwin\home\Administrator\buildbot\worker-tabularasa\tester_win64\build\share\julia\stdlib\v1.2\FileWatching\test\runtests.jl:208
  Expression: #= C:\cygwin\home\Administrator\buildbot\worker-tabularasa\tester_win64\build\share\julia\stdlib\v1.2\FileWatching\test\runtests.jl:208 =# @elapsed(#= C:\cygwin\home\Administrator\buildbot\worker-tabularasa\tester_win64\build\share\julia\stdlib\v1.2\FileWatching\test\runtests.jl:208 =# @test(watch_folder(dir, 0) == ("" => FileWatching.FileEvent()))) <= 0.5
   Evaluated: 1.334171801 <= 0.5

kmsquire · 2019-04-08T22:06:17Z

(Should be squashed on merge)

StefanKarpinski · 2019-04-09T14:51:49Z

@staticfloat: I can't tell if these test runs failing is related or not; also, does closing and opening the PR restart these CI runs like it does for other CI services?

staticfloat · 2019-04-09T17:44:05Z

Yes, it will. You can also click on the Details link, then click the rebuild button in the top right to re-run them.

The MacOS failure is the pbcopy issue, should be fixed on rebase.
The Win64 failure has a bunch of FileWatching failures where something that should have happened in <= 2 seconds instead happened in 60 seconds. I don't know what to make of that, but I'm willing to bet it's not this PR's fault.
The Win32 failure has some permissions errors; I've noticed this happening every now and then on the windows buildbots, I think it's because a previous Julia run stays resident in memory as a zombie process.

I am restarting the windows builds, but IMO this PR is good to go.

kmsquire · 2019-04-12T01:29:19Z

Rebased to address the pbcopy issue (#31653) .

The win64 FileWatching also failures appeared again before the rebase, so might look to see if that code calls into this codepath if it appears again.

kmsquire · 2019-04-12T14:00:31Z

The previous errors are gone, but at least one new one has cropped up: a failure while detecting ambiguities in LinearAlgebra on OSX. I did one restart, and the failure is the same.

~~There was also a new OSX Travis failure that I restarted, but hasn't completed yet.~~ Travis testing passed.

kmsquire · 2019-04-12T17:27:24Z

Repeated LinearAlgebra failure:

      From worker 5:	Error During Test at /Users/sabae/buildbot/worker-tabularasa/tester_macos64/build/share/julia/stdlib/v1.3/LinearAlgebra/test/ambiguous_exec.jl:4
      From worker 5:	  Test threw exception
      From worker 5:	  Expression: detect_ambiguities(LinearAlgebra; imported=true, recursive=true) == []
      From worker 5:	  TypeError: in TypeVar, in upper bound, expected Type, got Core.Compiler.UseRef
      From worker 5:	  Stacktrace:
      From worker 5:	   [1] typeintersect at ./reflection.jl:560 [inlined]
      From worker 5:	   [2] #isambiguous#23(::Bool, ::typeof(Base.isambiguous), ::Method, ::Method) at ./reflection.jl:1266
      From worker 5:	   [3] #isambiguous at ./none:0 [inlined]
      From worker 5:	   [4] #detect_ambiguities#30(::Bool, ::Bool, ::Bool, ::typeof(detect_ambiguities), ::Module) at /Users/julia/buildbot/worker/package_macos64/build/usr/share/julia/stdlib/v1.3/Test/src/Test.jl:1439
      From worker 5:	   [5] (::getfield(Test, Symbol("#kw##detect_ambiguities")))(::NamedTuple{(:imported, :recursive),Tuple{Bool,Bool}}, ::typeof(detect_ambiguities), ::Module) at ./none:0
      From worker 5:	   [6] top-level scope at /Users/sabae/buildbot/worker-tabularasa/tester_macos64/build/share/julia/stdlib/v1.3/LinearAlgebra/test/ambiguous_exec.jl:4
      From worker 5:	   [7] include at ./boot.jl:328 [inlined]
      From worker 5:	   [8] include_relative(::Module, ::String) at ./loading.jl:1094
      From worker 5:	   [9] include(::Module, ::String) at ./Base.jl:31
      From worker 5:	   [10] exec_options(::Base.JLOptions) at ./client.jl:295
      From worker 5:	   [11] _start() at ./client.jl:464
      From worker 5:	  
      From worker 5:	ERROR: Error while loading expression starting at /Users/sabae/buildbot/worker-tabularasa/tester_macos64/build/share/julia/stdlib/v1.3/LinearAlgebra/test/ambiguous_exec.jl:4
      From worker 5:	caused by [exception 1]
      From worker 5:	There was an error during testing
      From worker 5:	running testset LinearAlgebra/bunchkaufman...

@staticfloat, when restarting the buildbot tests, is the patch reapplied to the latest master, or is the branch compiled and run as is?

(In other words, is rebasing on master necessary when errors are fixed. My impression is yes, except maybe on AppVeyor?)

Edit: direct link to failed log: https://build.julialang.org/#/builders/69/builds/632

staticfloat · 2019-04-12T20:37:45Z

The buildbots just check out this branch; so yes you need to rebase to incorporate fixes from master.

kmsquire · 2019-04-12T22:23:30Z

Different error now in buildbot/package_linux32:

error during bootstrap:
LoadError("sysimg.jl", 16, LoadError("/buildworker/worker/package_linux32/build/usr/share/julia/stdlib/v1.3/LinearAlgebra/src/LinearAlgebra.jl", 355, LoadError("/buildworker/worker/package_linux32/build/usr/share/julia/stdlib/v1.3/LinearAlgebra/src/triangular.jl", 593, TypeError(:TypeVar, "upper bound", Type, Core.Compiler.UseRefIterator((Core.Compiler.UseRef(nothing, 0), nothing), false)))))
rec_backtrace at /buildworker/worker/package_linux32/build/src/stackwalk.c:94
record_backtrace at /buildworker/worker/package_linux32/build/src/task.c:210 [inlined]
jl_throw at /buildworker/worker/package_linux32/build/src/task.c:417
jl_parse_eval_all at /buildworker/worker/package_linux32/build/src/ast.c:895
jl_load at /buildworker/worker/package_linux32/build/src/toplevel.c:861
exec_program at /buildworker/worker/package_linux32/build/ui/repl.c:35
true_main at /buildworker/worker/package_linux32/build/ui/repl.c:108
main at /buildworker/worker/package_linux32/build/ui/repl.c:217
__libc_start_main at /build/glibc-PNN5fi/glibc-2.19/csu/libc-start.c:287
_start at /buildworker/worker/package_linux32/build/usr/bin/julia (unknown line)

Link: https://build.julialang.org/#builders/35/builds/1733

* Might also address JuliaLang#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

kmsquire · 2019-04-23T16:23:25Z

Tests have all passed now. Mergeable?

kmsquire · 2019-04-26T14:55:52Z

Bump.

kmsquire force-pushed the kms/fix-search-sorted branch 3 times, most recently from 1179a40 to b0165eb Compare April 6, 2019 19:34

kmsquire changed the title ~~Loosen searchsorted* index type (fixes #30763)~~ WIP: Loosen searchsorted* index type (fixes #30763) Apr 8, 2019

StefanKarpinski approved these changes Apr 8, 2019

View reviewed changes

kmsquire changed the title ~~WIP: Loosen searchsorted* index type (fixes #30763)~~ Loosen searchsorted* index type (fixes #30763) Apr 8, 2019

kmsquire force-pushed the kms/fix-search-sorted branch from 93c5ac4 to 10ff01a Compare April 12, 2019 01:25

kmsquire force-pushed the kms/fix-search-sorted branch from 10ff01a to ae48836 Compare April 12, 2019 21:59

kmsquire mentioned this pull request Apr 12, 2019

Partially revert "Remove specialized sort! method for PartialQuickSort{Int} (fixes #12833)" #31632

Merged

kmsquire mentioned this pull request Apr 12, 2019

Random buildbot failures detecting ambiguities in LinearAlgebra #31712

Closed

Loosen searchsorted* index type (fixes JuliaLang#30763, JuliaLang#31618)

c2e330b

* Might also address JuliaLang#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

kmsquire force-pushed the kms/fix-search-sorted branch from ae48836 to c2e330b Compare April 17, 2019 06:33

StefanKarpinski merged commit 458380c into JuliaLang:master Apr 29, 2019

abhinavmehndiratta mentioned this pull request May 2, 2019

searchsortedfirst fails with non-Int indices #31618

Closed

simonschoelly mentioned this pull request Jun 23, 2019

Fix adjacency matrix for all types sbromberger/LightGraphs.jl#1170

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Loosen searchsorted* index type (fixes #30763) #31633

Loosen searchsorted* index type (fixes #30763) #31633

kmsquire commented Apr 6, 2019

kmsquire commented Apr 8, 2019

kmsquire commented Apr 8, 2019

kmsquire commented Apr 8, 2019

StefanKarpinski left a comment

kmsquire commented Apr 8, 2019

kmsquire commented Apr 8, 2019

StefanKarpinski commented Apr 9, 2019

staticfloat commented Apr 9, 2019

kmsquire commented Apr 12, 2019

kmsquire commented Apr 12, 2019 •

edited

Loading

kmsquire commented Apr 12, 2019 •

edited

Loading

staticfloat commented Apr 12, 2019

kmsquire commented Apr 12, 2019

kmsquire commented Apr 23, 2019

kmsquire commented Apr 26, 2019

Loosen searchsorted* index type (fixes #30763) #31633

Loosen searchsorted* index type (fixes #30763) #31633

Conversation

kmsquire commented Apr 6, 2019

kmsquire commented Apr 8, 2019

kmsquire commented Apr 8, 2019

kmsquire commented Apr 8, 2019

StefanKarpinski left a comment

Choose a reason for hiding this comment

kmsquire commented Apr 8, 2019

kmsquire commented Apr 8, 2019

StefanKarpinski commented Apr 9, 2019

staticfloat commented Apr 9, 2019

kmsquire commented Apr 12, 2019

kmsquire commented Apr 12, 2019 • edited Loading

kmsquire commented Apr 12, 2019 • edited Loading

staticfloat commented Apr 12, 2019

kmsquire commented Apr 12, 2019

kmsquire commented Apr 23, 2019

kmsquire commented Apr 26, 2019

kmsquire commented Apr 12, 2019 •

edited

Loading

kmsquire commented Apr 12, 2019 •

edited

Loading