I keep getting worldage errors

General Usage
1

Last few days I’ve been getting random world age errors. Here is the code that triggers it now (the last line).

using DataFrames
using CSV
using GLM
using RCall
using Plots
#using PyPlot
using Mamba 
using StatsPlots

dat = CSV.read("./regression_models/Leinhardt.csv")
cols = Symbol.(names(dat))

(nobs, nvars) = size(dat)
l = @layout grid(4,4)
plots = Plots.Plot{Plots.GRBackend}[]

# Plot data for visual inspection
for i = 1:nvars
    for j = 1:nvars
        if i != j
            p = @df dat scatter(cols(cols[i]), cols(cols[j]))
            push!(plots, p)
        else
            p = @df dat scatter(cols(cols[i]), cols(cols[j]))
            push!(plots, p)
        end
    end
end
Plots.plot(plots..., layout=l, legend=false, size=(1200, 800))

## data is very right skewed, so lets take the log of the continous variables. 
@. dat[!, :logincome] = log(dat[:, :income])
@. dat[!, :loginfant] = log(dat[:, :infant]) ## TRIGGERED HERE

The error is

julia> @. dat[!, :loginfant] = log(dat[:, :infant])
105-element SentinelArrays.SentinelArrayInternal Error: {ERROR: Float64MethodError: no method matching size(::SentinelArrays.SentinelArray{Float64,1,Float64,Missing,Array{Float64,1}})
The applicable method may be too new: running in world age 27138, while current world is 27283.

Ran the same line again and got

julia> Dict{String,Any}) at /home/affan/.vscode-server/extensions/julialang.language-julia-0.16.8/scripts/packages/JSONRPC/src/typed.jl:63
 [5] macro expansion at /home/affan/.vscode-server/extensions/julialang.language-julia-0.16.8/scripts/packages/VSCodeServer/src/VSCodeServer.jl:83 [inlined]
 [6] (::VSCodeServer.var"#53#55"{Bool,String})() at ./task.jl:358

and it just stuck there until I hit enter a few times.

Using VS Code.

2

This could be a variant of https://github.com/julia-vscode/julia-vscode/issues/1338. I’ve been hitting this one a lot since the latest update as well.

3

We hope to push an update out tomorrow that fixes that, just need a few more hours on the insider channel with any major regressions.

1 Like