Add support for Julia v0.7

[juliohm]

Could you please help fix this build error in Julia v0.7? I am not sure how to fix it myself.

ERROR: LoadError: LoadError: TypeError: in SeriesAnnotations, in type definition, expected Type, got typeof(Base.Nullable)
Stacktrace:
 [1] top-level scope
 [2] include at ./boot.jl:314 [inlined]
 [3] include_relative(::Module, ::String) at ./loading.jl:1067
 [4] include at ./sysimg.jl:29 [inlined]
 [5] include(::String) at /home/juliohm/.julia/v0.7/Plots/src/Plots.jl:3
 [6] top-level scope
 [7] include at ./boot.jl:314 [inlined]
 [8] include_relative(::Module, ::String) at ./loading.jl:1067
 [9] include(::Module, ::String) at ./sysimg.jl:29
 [10] top-level scope
 [11] eval at ./boot.jl:317 [inlined]
 [12] top-level scope at ./<missing>:3
in expression starting at /home/juliohm/.julia/v0.7/Plots/src/components.jl:438
in expression starting at /home/juliohm/.julia/v0.7/Plots/src/Plots.jl:157
ERROR: Failed to precompile Plots to /home/juliohm/.julia/compiled/v0.7/Plots.ji.

Please let me know if there is a branch with support added.

[mkborregaard]

This is due to JuliaLang/julia#23642 i.e. the deprecation of Nullable. The fix would seem to be to remove Nullable's throughout on a 0.7-only branch. I've just made one.

When will it be possible to use femtocleaner to do these things automatically?

[juliohm]

Thank you @mkborregaard! I checked the branch locally, but still getting the same error. Sorry for the delay to reply, I was in a conference these two last days. Coding during the coffee breaks...

[mkborregaard]

Oh I didn't actually do anything I just created the branch. Perhaps it's as easy as find-replacing Nullable{SomeType} to Union{SomeType, Nothing} - I've tried that in #1508 , which you could try to check out?

[juliohm]

I pulled the julia0.7 branch, and patched the last modifications you linked in the PR, but didn't have much success. I will try to export some data and plot Julia v0.6 meanwhile just to get things going during this migration.

[mkborregaard]

OK, thanks for giving it a shot

[ederag]

What about using the Nullable package https://github.com/JuliaArchive/Nullables.jl ?

[mkborregaard]

No, rather look forward. That's a legacy package.

[ederag]

Sorry, I now see that it was not a mere cleanup of Base, but rather a true deprecation, because of performance issues.
Little progress:
mkborregaard#2
JuliaPlots/PlotUtils.jl#39
Also needs JuliaGraphics/Showoff.jl#10 until it is merged.

[RalphAS]

I have a branch (ras/julia0.7) in my fork (https://github.com/RalphAS/Plots.jl) running under 0.7-alpha, with gr and pyplot backends on Linux. (I am using the julia0.7 branches of PlotUtils and RecipesBase too.) Since I am working in v0.7 the package management is shaky, so I'm not sure how best to make this useful to the main project. Can you cherry-pick commits from a fork, or would you like me to attempt a WIP PR? (looking for advice from @mkborregaard a/o @daschw)

[vishaluk]

I got the Julia 0.7 rc2 installed and I am unable to use Plots at all because it gives me the above mentioned error. Is this still an open issue?

[JeffFessler]

I just installed the brand new official Julia 0.7 featured on the Julia homepage and got this same error. It is quite ironic because the "Plots" package is the first one highlighted on Julia's home page! It is definitely still an open issue!

[sjmgarnier]

Same problem as @JeffFessler using the new official Julia 0.7.

[mkborregaard]

You're right - the solution for now is ]add Plots#master but there should be a 0.7 compatible release.
Here's the reason there isn't one yet: Plots has a very extensive test suite that involves quite a lot of packages (that are not dependencies of Plots). If just one of these packages is not yet 0.7 ready, Travis CI will fail for Plots though in fact Plots generally works perfectly - and you can't release a package with failing tests. We're on it.

[mkborregaard]

JuliaLang/METADATA.jl#16449

[JeffFessler]

With both Julia 0.7.0 and 1.0.0, Plots v0.19.2 seems to be working fine now at least with default gr() backend. Thanks!