Tuple not displaying properly in REPL

[wheineman]

$ julia
               _
   _       _ _(_)_     |  A fresh approach to technical computing
  (_)     | (_) (_)    |  Documentation: http://docs.julialang.org
   _ _   _| |_  __ _   |  Type "help()" to list help topics
  | | | | | | |/ _` |  |
  | | |_| | | | (_| |  |  Version 0.3.0-prerelease+3908 (2014-06-26 23:41 UTC)
 _/ |\__'_|_|_|\__'_|  |  Commit 8f99257 (0 days old master)
|__/                   |  x86_64-w64-mingw32

julia> x = tuple(1,2,3)
(1,2,0)

julia> x
(1,2,0)

julia> x[3]
3

julia>
julia> versioninfo()
Julia Version 0.3.0-prerelease+3908
Commit 8f99257 (2014-06-26 23:41 UTC)
Platform Info:
  System: Windows (x86_64-w64-mingw32)
  CPU: Intel(R) Core(TM) i7-4600M CPU @ 2.90GHz
  WORD_SIZE: 64
  BLAS: libopenblas (USE64BITINT DYNAMIC_ARCH NO_AFFINITY)
  LAPACK: libopenblas
  LIBM: libopenlibm

julia>

[quinnj]

Can you post the output of versioninfo()?

[adamkapor]

I'm having the same problem. This is only on my windows machine, not on the mac.

julia> versioninfo()
Julia Version 0.3.0-prerelease+3907
Commit ac807db (2014-06-26 17:32 UTC)
Platform Info:
  System: Windows (x86_64-w64-mingw32)
  CPU: Intel(R) Core(TM) i7-3930K CPU @ 3.20GHz
  WORD_SIZE: 64
  BLAS: libopenblas (USE64BITINT DYNAMIC_ARCH NO_AFFINITY)
  LAPACK: libopenblas
  LIBM: libopenlibm

[adamkapor]

               _
   _       _ _(_)_     |  A fresh approach to technical computing
  (_)     | (_) (_)    |  Documentation: http://docs.julialang.org
   _ _   _| |_  __ _   |  Type "help()" to list help topics
  | | | | | | |/ _` |  |
  | | |_| | | | (_| |  |  Version 0.3.0-prerelease+3907 (2014-06-26 17:32 UTC)
 _/ |\__'_|_|_|\__'_|  |  Commit ac807db (0 days old master)
|__/                   |  x86_64-w64-mingw32

julia> b=(1,2,3)
(1,2,0)

julia> b=(1,2,3,4)
(1,2,0,0)

julia> print(b)
(1,2,3,4)
julia> show(b)
(1,2,3,4)
julia> b
(1,2,0,0)

[quinnj]

I see this at the REPL, but not in IJulia. @stevengj, any ideas?

[mbauman]

Looks like this is architecture-specific, too. I don't see it on my Sandy Bridge Xeon E31235 (x86_64-w64-mingw32).

[stevengj]

I can't reproduce on MacOS in either the REPL or IJulia.

Note that both the REPL and IJulia use the same writemime(io, "text/plain", x) to get their display text for tuples, so I don't see why there would be a difference. Are you sure they are running the same Julia version?

[quinnj]

Hmmm, I can't seem to reproduce with show, writemime, etc. What am I missing here?

julia> writemime(STDOUT,MIME("text/plain"),(1,1,1))
(1,1,1)
julia> show(STDOUT,(1,1,1))
(1,1,1)
julia> writemime(STDOUT,MIME("text/plain"),(1,1,1))
(1,1,1)
julia> (1,1,1)
(1,1,0)

[wheineman]

I suspect this is Windows REPL only, and maybe only Windows REPL only on certain architectures. I see it on my Haswell machines.

[ihnorton]

@Keno @vtjnash this appears to be architecture-specific.

Using latest win64 nightly:

• not present on Westmere (Xeon E5620)
• present on Ivy Bridge (i7-3940XM)

Using recent win32, and 0.2.1

• not present on Ivy Bridge

[Keno]

Tried with LLVM 3.5 (I can reproduce with 3.3) and the problem went away. So I'm gonna assume this is fixed in 3.5 and not try to track this down.

[KrisKusano]

Looks like this one is pretty closed down, but I can reproduce this in the newest 0.3.0 version (Intel i7-4700MQ, Haswell, Windows 7 64).

julia> a = (1, 5, 9)
(1,5,0)

julia> a[end]
9

julia> versioninfo()
Julia Version 0.3.0
Commit 7681878 (2014-08-20 20:43 UTC)
Platform Info:
  System: Windows (x86_64-w64-mingw32)
  CPU: Intel(R) Core(TM) i7-4700MQ CPU @ 2.40GHz
  WORD_SIZE: 64
  BLAS: libopenblas (USE64BITINT DYNAMIC_ARCH NO_AFFINITY Haswell)
  LAPACK: libopenblas
  LIBM: libopenlibm
  LLVM: libLLVM-3.3

[iamed2]

I was able to reproduce with generic but not specialized display().

julia> b = (1,2,3)
(1,2,0)

julia> display(b)
(1,2,0)

julia> display(Base.Multimedia.displays[1], b)
(1,2,3)
julia> display(Base.Multimedia.displays[2], b)
(1,2,3)

Note: displays[2] corresponds to the REPL display used in the first two examples.

[nalimilan]

This keeps being reported. Maybe worth looking for a fix? Or do you want to switch to LLVM 3.5 for 0.3.1 (risky...)?

[DumpsterDoofus]

I observed this too, using Ivy Bridge Intel Core i5-3550 on 64-bit Windows 7 on Julia 0.3.0 64-bit. Glad to hear it isn't affecting the actual computations, although it was a bit amusing when I first noticed it.

[tkelman]

There are test failures on win64 with LLVM 3.5.0, so I don't think this display problem is worth switching for just yet.

[IainNZ]

Another case on SO: http://stackoverflow.com/questions/28423504/why-does-julia-1-2-3-return-1-2-0/28423717#28423717

[adijkstra]

I also still have this issue in 0.3.5

julia> A = (1,2,3)
(1,2,0)

julia> A
(1,2,0)

julia> versioninfo()
Julia Version 0.3.5
Commit a05f87b* (2015-01-08 22:33 UTC)
Platform Info:
System: Windows (x86_64-w64-mingw32)
CPU: Intel(R) Core(TM) i3-3220 CPU @ 3.30GHz
WORD_SIZE: 64
BLAS: libopenblas (USE64BITINT DYNAMIC_ARCH NO_AFFINITY Sandybridge)
LAPACK: libopenblas
LIBM: libopenlibm
LLVM: libLLVM-3.3

[pao]

@adijkstra, that is expected; we are unlikely to be updating the LLVM dependency to 3.5 or later, where the problem is fixed, in the 0.3.x series of Julia releases due to extensive changes in LLVM's APIs. Note from your versioninfo() output that your build is using LLVM 3.3.

[tkelman]

Given that it's looking like we'll also have to stay on LLVM 3.3 for Julia 0.4.0, would it be worth digging into exactly where in the show pipeline this happens? I unfortunately don't have reliable access to a machine where this happens.