Julia + Centos - Arpack cannot be built - centos

Question

I installed julia (v1.1.1) via as stated in [platform specific instructions(https://julialang.org/downloads/platform.html) on my CentOS machine using yum and epel.
Julia works fine, but when precompiling packages (using DataFrames for example), I get
[ Info: Precompiling GLM [38e38edf-8417-5370-95a0-9cbb8c7f171a]
ERROR: LoadError: No deps.jl file could be found. Please try running Pkg.build("Arpack").
Currently, the build command might fail when Julia has been built from source
and the recommendation is to use the official binaries from julialang.org.
For more info see https://github.com/JuliaLinearAlgebra/Arpack.jl/issues/5.
I don't know enough about compilations, building from source, etc. to solve this. I guess it would work fine if I just download Julia from the main page and use it from within that folder. But I would prefer using it via package manager to automatically receive updates.
How do I solve this?

Answer 1

Unfortunately this is expected since Arpack.jl expects OpenBLAS to be called libopenblas64_.so.0, while in Fedora/CentOS it is libopenblasp64_.so.0. You can work around the problem using sudo ln -s /lib64/libopenblas64p_.so.0 ~/.julia/packages/Arpack/*/deps/lib/libopenblas64_.so.0. I'll try to find a solution with Arpack.jl developers.

Answer 2

So I eventually solved this via help of the original julialang page. For some reason, Arpack can't be built if you install julia via # yum install julia (with epel enabled). Pkg.build("Arpack") fails.
If you download julia from the official page and add it to path, it still throws the same error message. If you however start up julia and run Pkg.build("Arpack") beforehand, it works fine!

Answer 3

I confirm that #MilanBouchet-Valat solution works. Here is some details for Debian distribution.
I have installed Julia using Debian 4.19.37-5 package manager (synaptic for instance). I have
julia> versioninfo()
prints
Julia Version 1.0.4
Platform Info:
OS: Linux (x86_64-linux-gnu)
CPU: Intel(R) Xeon(R) CPU E5-2603 v3 # 1.60GHz
WORD_SIZE: 64
LIBM: libopenlibm
LLVM: libLLVM-6.0.0 (ORCJIT, haswell)
The symptoms
Under Julia I got the "expected" error when trying to install the Arpack package:
(v1.0) pkg> add Arpack
prints
Resolving package versions...
Updating `~/.julia/environments/v1.0/Project.toml`
[7d9fca2a] + Arpack v0.3.1
Updating `~/.julia/environments/v1.0/Manifest.toml`
[no changes]
then
julia> using Arpack
prints
[ Info: Precompiling Arpack [7d9fca2a-8960-54d3-9f78-7d1dccf2cb97]
ERROR: LoadError: No deps.jl file could be found. Please try running
Pkg.build("Arpack").
Currently, the build command might fail when Julia has been built from source
and the recommendation is to use the official binaries from julialang.org.
For more info see https://github.com/JuliaLinearAlgebra/Arpack.jl/issues/5.
Stacktrace:
[1] top-level scope at /home/picaud/.julia/packages/Arpack/cu5By/src/Arpack.jl:19
[2] include at ./boot.jl:317 [inlined]
[3] include_relative(::Module, ::String) at ./loading.jl:1044
[4] include(::Module, ::String) at ./sysimg.jl:29
[5] top-level scope at none:2
[6] eval at ./boot.jl:319 [inlined]
[7] eval(::Expr) at ./client.jl:393
[8] top-level scope at ./none:3
in expression starting at /home/picaud/.julia/packages/Arpack/cu5By/src/Arpack.jl:16
ERROR: Failed to precompile Arpack [7d9fca2a-8960-54d3-9f78-7d1dccf2cb97] to /home/picaud/.julia/compiled/v1.0/Arpack/X5VZL.ji.
Stacktrace:
[1] error(::String) at ./error.jl:33
[2] compilecache(::Base.PkgId, ::String) at ./loading.jl:1203
[3] _require(::Base.PkgId) at ./loading.jl:960
[4] require(::Base.PkgId) at ./loading.jl:858
[5] require(::Module, ::Symbol) at ./loading.jl:853
however
pkg> build Arpack
prints
Building Arpack → `~/.julia/packages/Arpack/cu5By/deps/build.log`
┌ Error: Error building `Arpack`:
│ ERROR: LoadError: LibraryProduct(nothing, ["libarpack"], :libarpack, "Prefix(/home/picaud/.julia/packages/Arpack/cu5By/deps/usr)") is not satisfied, cannot generate deps.jl!
│ Stacktrace:
│ [1] error(::String) at ./error.jl:33
│ [2] #write_deps_file#152(::Bool, ::Function, ::String, ::Array{LibraryProduct,1}) at /home/picaud/.julia/packages/BinaryProvider/TcAwt/src/Products.jl:414
│ [3] (::getfield(BinaryProvider, Symbol("#kw##write_deps_file")))(::NamedTuple{(:verbose,),Tuple{Bool}}, ::typeof(write_deps_file), ::String, ::Array{LibraryProduct,1}) at ./none:0
│ [4] top-level scope at none:0
│ [5] include at ./boot.jl:317 [inlined]
│ [6] include_relative(::Module, ::String) at ./loading.jl:1044
│ [7] include(::Module, ::String) at ./sysimg.jl:29
│ [8] include(::String) at ./client.jl:392
│ [9] top-level scope at none:0
│ in expression starting at /home/picaud/.julia/packages/Arpack/cu5By/deps/build.jl:74
└ # Pkg.Operations /build/julia-gbfU8y/julia-1.0.4+dfsg/usr/share/julia/stdlib/v1.0/Pkg/src/Operations.jl:1096
The diagnostic
You can check that this is a dependance problem with:
ldd ~/.julia/packages/Arpack/cu5By/deps/usr/lib/libarpack.so
prints
linux-vdso.so.1 (0x00007ffc179a1000)
libopenblas64_.so.0 => not found <----
libgfortran.so.5 => /usr/lib/x86_64-linux-gnu/libgfortran.so.5 (0x00007feea073d000)
libm.so.6 => /lib/x86_64-linux-gnu/libm.so.6 (0x00007feea05ba000)
libgcc_s.so.1 => /lib/x86_64-linux-gnu/libgcc_s.so.1 (0x00007feea05a0000)
libquadmath.so.0 => /usr/lib/x86_64-linux-gnu/libquadmath.so.0 (0x00007feea0557000)
libc.so.6 => /lib/x86_64-linux-gnu/libc.so.6 (0x00007feea0396000)
/lib64/ld-linux-x86-64.so.2 (0x00007feea0c67000)
Debian distro:
find /usr/lib/ -name libopenblas\*
prints
/usr/lib/x86_64-linux-gnu/libopenblas.so
/usr/lib/x86_64-linux-gnu/libopenblasp-r0.3.6.a
/usr/lib/x86_64-linux-gnu/libopenblas64_.so.0
/usr/lib/x86_64-linux-gnu/julia/libopenblas.so
/usr/lib/x86_64-linux-gnu/libopenblas.so.0
/usr/lib/x86_64-linux-gnu/libopenblas.a
/usr/lib/x86_64-linux-gnu/libopenblasp-r0.3.6.so
The "fix"
With root privileges (su) create a symbolic links
ln -s /usr/lib/x86_64-linux-gnu/libopenblas.so /usr/lib/x86_64-linux-gnu/libopenblas64_.so.0
Now it works
pkg> build Arpack
Building Arpack → `~/.julia/packages/Arpack/cu5By/deps/build.log`
then
using Arpack
[ Info: Precompiling Arpack [7d9fca2a-8960-54d3-9f78-7d1dccf2cb97]
are OK