I hope it’s OK to repost UndefVarError: Mat not defined issue here.
Should I try another version of Julia to solve it? Is Windows more problematic for visualization with Julia than Linux?
I hope it’s OK to repost UndefVarError: Mat not defined issue here.
Should I try another version of Julia to solve it? Is Windows more problematic for visualization with Julia than Linux?
Sorry about that, I still haven’t tagged a version that works on 0.6!
Please try Julia 0.5 for now!
Best,
Simon
Thank you, Simon. I installed Julia 0.5.2 on Ubuntu 16.04 and GLVisualize passes the tests. I still have issues on Windows 7, though. I installed JuliaPro 0.5.2 and I’m getting this error on Pkg.add("GLVisualize)
:
LoadError: could not spawn `'C:\Users\julia\AppData\Local\Julia-0.5.2\bin\7z.exe' x 'C:\JuliaPro-0.5.2.2\pkgs-0.5.2.2\v0
.5\GLFW\deps\downloads\glfw-3.2.1.bin.WIN64.zip' -y '-oC:\JuliaPro-0.5.2.2\pkgs-0.5.2.2\v0.5\GLFW\deps'`: no such file o
r directory (ENOENT)
while loading C:\JuliaPro-0.5.2.2\pkgs-0.5.2.2\v0.5\GLFW\deps\build.jl, in expression starting on line 52
I don’t know why the build process is looking into C:\Users\julia\AppData\Local\Julia-0.5.2\bin
directory, which doesn’t exist.
Maybe the Julia Pro install doesn’t ship with 7zip?
Or something went wrong with the download of the binary dependency!
Delete C:\JuliaPro-0.5.2.2\pkgs-0.5.2.2\lib\v0.5\BinDeps.ji
and try again
Thanks. Removing this file and rebuilding GLFW allows for GLVisualize test to run and complete almost successfully, except ffmpeg, which I don’t need.
BTW, I’m getting this message while running GLVisualize test:
WARNING: Plots.jl is not installed, excluding a lot of interesting examples from the tests
and a nice summary.
Please consider doing: `Pkg.add("Plots"); Pkg.checkout("Plots", "dev")`
I did Pkg.add("Plots"); Pkg.checkout("Plots", "dev")
and I checked it with Pkg.status("Plots")
, but it still insists that Plots.jl is not installed.
Yes, that’s a “bug” - it doesn’t mean anything!
It should actually work with the currently tagged Plots.jl version!
Speaking of tagging. What is the way to find out what versions of Julia are supported by a package?
The reason I was trying GLVisualize with Julia 0.6 is that Julia packages page shows the same untestable
status for both 0.5 and 0.6 versions.
If you go to a package’s repository, you’ll see badges at the top. For this package:
there’s just one for tests: “Build status”, which leads to:
https://ci.maleadt.net/buildbot/julia/builders/GLVisualize.jl%3A%20Julia%200.5%20(x86-64)
This shows it’s only testing on v0.5.
Notice that GLVisualize is an outlier here. Most packages will test on Travis and AppVeyor. Let’s pick a random package:
At the top you see the “Build Status” with badges for Travis and AppVeyor. If you click on them, you’ll get:
https://ci.appveyor.com/project/blegat/optim-jl
Looks like Travis had a random failure on Mac, but it’s working on Linux v0.6 so as of now it’s testing on v0.6+. Meanwhile, the AppVeyor tests (Windows) haven’t ran for 5 months and are stuck at v0.5, so this needs a PR (I might do it if I remember today).
So why does GLVisualize do something non-standard? My guess is because it needs a buildbot which has GPUs to test with. And that’s probably why it’s disabled for PkgEval. I can confirm this by looking at the page for the buildbot:
https://ci.maleadt.net/buildbot/julia/buildslaves/hydor.elis.ugent.be
host: Linux x64 running Debian stable, 2x E5-2637 with 64GB RAM, NVIDIA TITAN.
so this is only CI tested on Linux, but it is tested with an NVIDIA Titan.
Thanks a lot. Visualization packages present some testing difficulties, as you noticed. OpenGL drivers and other GPU related infrastructure is required.