using QuantEcon failing on v0.4

[ZacCranko]

Anyone know why this is happening with the master branch on v0.4?

julia> using QuantEcon
WARNING: requiring "Distributions" did not define a corresponding module.
WARNING: requiring "Distributions" did not define a corresponding module.
WARNING: requiring "Distributions" did not define a corresponding module.
WARNING: module Util should explicitly import * from Base
WARNING: module Filters should explicitly import * from Base
WARNING: requiring "Distributions" did not define a corresponding module.
WARNING: requiring "Distributions" did not define a corresponding module.
ERROR: LoadError: LoadError: UndefVarError: MultivariateNormal not defined
 in include at /usr/local/Cellar/julia/HEAD/lib/julia/sys.dylib
 in include_from_node1 at /usr/local/Cellar/julia/HEAD/lib/julia/sys.dylib
 in include at /usr/local/Cellar/julia/HEAD/lib/julia/sys.dylib
 in include_from_node1 at /usr/local/Cellar/julia/HEAD/lib/julia/sys.dylib
 in require at /usr/local/Cellar/julia/HEAD/lib/julia/sys.dylib
while loading /Users/zaccranko/.julia/v0.4/QuantEcon/src/lss.jl, in expression starting on line 41
while loading /Users/zaccranko/.julia/v0.4/QuantEcon/src/QuantEcon.jl, in expression starting on line 120

[sglyon]

Will you give me the output of versioninfo()

[ZacCranko]

julia> versioninfo()
Julia Version 0.4.0-dev+6642
Commit 58dc82f* (2015-08-11 23:19 UTC)
Platform Info:
  System: Darwin (x86_64-apple-darwin15.0.0)
  CPU: Intel(R) Core(TM) i7-4870HQ CPU @ 2.50GHz
  WORD_SIZE: 64
  BLAS: libopenblas (NO_AFFINITY HASWELL)
  LAPACK: libopenblas
  LIBM: libopenlibm
  LLVM: libLLVM-3.3

[sglyon]

I haven't tried on today's master.

Can you load distributions by itself?

[ZacCranko]

I did try that, Distributions loads without complaint

[sglyon]

Hmm.

Have you by chance tried using Base.compile with these packages?

If so maybe try deleting the cache files...

[ZacCranko]

I haven't, this is a completely clean build of Julia that i haven't messed with. Is Travis building v0.4?

[sglyon]

I don't think we've pushed to master recently enough. Usually nightly builds are a little behind anyway.

That's is quite strange. I guess you could try the nuclear option and delete your whole package directory... I don't have great advice for this thing

[ZacCranko]

Okay so just did rm -rf ~/.julia/v0.4 then Pkg.add("QuantEcon"), bad news still getting the same issue

julia> using QuantEcon
WARNING: requiring "Distributions" did not define a corresponding module.
WARNING: requiring "Distributions" did not define a corresponding module.
WARNING: requiring "Distributions" did not define a corresponding module.
WARNING: module Util should explicitly import * from Base
INFO: Recompiling stale cache file /Users/zaccranko/.julia/lib/v0.4/Polynomials.ji for module Polynomials.
WARNING: module Filters should explicitly import * from Base
INFO: Recompiling stale cache file /Users/zaccranko/.julia/lib/v0.4/Docile.ji for module Docile.
INFO: Recompiling stale cache file /Users/zaccranko/.julia/lib/v0.4/LightGraphs.ji for module LightGraphs.
INFO: Recompiling stale cache file /Users/zaccranko/.julia/lib/v0.4/GZip.ji for module GZip.
INFO: Recompiling stale cache file /Users/zaccranko/.julia/lib/v0.4/StatsBase.ji for module StatsBase.
INFO: Recompiling stale cache file /Users/zaccranko/.julia/lib/v0.4/ArrayViews.ji for module ArrayViews.
INFO: Recompiling stale cache file /Users/zaccranko/.julia/lib/v0.4/StatsFuns.ji for module StatsFuns.
INFO: Recompiling stale cache file /Users/zaccranko/.julia/lib/v0.4/LightXML.ji for module LightXML.
INFO: Recompiling stale cache file /Users/zaccranko/.julia/lib/v0.4/ParserCombinator.ji for module ParserCombinator.
INFO: Recompiling stale cache file /Users/zaccranko/.julia/lib/v0.4/AutoHashEquals.ji for module AutoHashEquals.
WARNING: requiring "Distributions" did not define a corresponding module.
WARNING: requiring "Distributions" did not define a corresponding module.
ERROR: LoadError: LoadError: UndefVarError: MultivariateNormal not defined
 in include at /usr/local/Cellar/julia/HEAD/lib/julia/sys.dylib
 in include_from_node1 at /usr/local/Cellar/julia/HEAD/lib/julia/sys.dylib
 in include at /usr/local/Cellar/julia/HEAD/lib/julia/sys.dylib
 in include_from_node1 at /usr/local/Cellar/julia/HEAD/lib/julia/sys.dylib
 in require at /usr/local/Cellar/julia/HEAD/lib/julia/sys.dylib
while loading /Users/zaccranko/.julia/v0.4/QuantEcon/src/lss.jl, in expression starting on line 41
while loading /Users/zaccranko/.julia/v0.4/QuantEcon/src/QuantEcon.jl, in expression starting on line 120

[sglyon]

Hmm not sure.

I tried scanning the commit log for Julia for today and didn't notice anything that would explain this.

Unfortunately I need my Julia to be fully working to do my research so I can't really hop on the latest master to debug with you.

I suppose I could do it within a docker container or something, do you reckon it's worth it?

[ZacCranko]

I was able to reproduce it with the binary off julialang.org: https://status.julialang.org/download/osx10.7+
so you could grab that and see if its happening for you. Either way is this registering on travis? I can't tell.

[sglyon]

Yep. I'll try pretty soon.

We haven't pushed to master to register a build on Travis so I'm not sure

[sglyon]

I did have the same problems with the binary.

Doing

rm -rf ~/.julia/lib

Then

julia> Base.compilecache(:Distributions)
"/Users/sglyon/.julia/lib/v0.4/Distributions.ji"

julia> Base.compilecache(:QuantEcon)

fixed it for me.

I did have to move to the master version of Docile.jl and I was already on the master of distributions (not sure if being on Distributions master is necessary).

[ZacCranko]

What was the problem?

[sglyon]

not sure...

Somehow forcing a cached version of Distributions and allowing using QuantEcon to call that instead of load at "using-time" made it work.

Can you verify that it fixes your issue?

[ZacCranko]

Okay fixed. So rm -rf ~/.julia/lib followed by Base.compilecache(:Distributions) did the trick. Base.compilecache(:QuantEcon) was not necessary. It's slightly concerning that this is necessary, I bet it's Distributions.jl's fault.

[sglyon]

Not sure... Just sounds like life on the bleeding edge to me

[vtjnash]

This is JuliaLang/julia#12586

[tkelman]

slash 9079 slash 4600... we need to fix that