0.7 updates

[quinnj]

No description provided.

[ararslan]

This seems like a dangerous definition to have lying around, but I guess it's okay until we stop supporting 0.6.

[nalimilan]

Yeah. Shouldn't Compat provide something like that, since many packages will have the same problem?

BTW, is Femtocleaner able to remove this kind of isdefined check? Or should we use VERSION checks instead?

[ararslan]

I'm not sure about FemtoCleaner, but pairs will be available in Compat as of 0.40.0: JuliaLang/METADATA.jl#12530

[nalimilan]

So do we want to use Compat.pairs here rather than a custom method?

[ararslan]

I'd say so, now that it's available in Compat. Just have to bump the REQUIRE to 0.40.0.

[quinnj]

now we might as well wait till Compat gets Unicode since DataFrames show totally breaks w/o it.

[ararslan]

Nah, I'd say just add

if VERSION >= v"0.7.0-DEV.2915"
    using Unicode
end

[ararslan]

WOW that 0.7 failure. 😱 Reported at JuliaLang/julia#25176

[bkamins]

I believe that this line:
cbb54c7#diff-0aa1ff52d56badabf7d1511d3dc3f279R119
is not working correctly with Julia 0.6 and all DataFrames tests will fail:

julia> using Compat.pairs

julia> function test(; kwargs...)
           collect((k => v for (k,v) in pairs(kwargs)))
       end
test (generic function with 1 method)

julia> test(a=1,b=2)
2-element Array{Pair{Int64,Tuple{Symbol,Int64}},1}:
 1=>(:a, 1)
 2=>(:b, 2)

or I am missing something?

[nalimilan]

Did I approve the changes despite CI not passing? I'm too used to seeing 0.7 fail, I didn't even check whether 0.6 passed...

[ararslan]

Yeah 0.6 was failing; I was surprised this got merged as-is.

[nalimilan]

See #1329.