Drop support for Julia 0.7

[hyrodium]

I think we can drop support for Julia 0.7.

[codecov]

Codecov Report

Merging #92 (362a589) into master (8f71f21) will increase coverage by 8.08%.
The diff coverage is n/a.

@@            Coverage Diff             @@
##           master      #92      +/-   ##
==========================================
+ Coverage   91.11%   99.19%   +8.08%     
==========================================
  Files           3        3              
  Lines         270      248      -22     
==========================================
  Hits          246      246              
+ Misses         24        2      -22     

Impacted Files	Coverage Δ
src/IntervalSets.jl	99.08% <ø> (+16.63%)	⬆️
src/interval.jl	99.09% <ø> (ø)

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 8f71f21...362a589. Read the comment docs.

[timholy]

I think this will require a minor version bump, although it's possible that Julia 0.7 is exempted.

[hyrodium]

I think dropping support is not a breaking change and it's okay with updating the patch version.

[hyrodium]

If we update the minor version bump (with major version 0), I'd like to drop the deprecated methods such as in(a::AbstractInterval, b::TypedEndpointsInterval{:closed,:closed}).

[timholy]

That might be true. I know that bumping the Julia version typically requires a minor version bump, but perhaps that's only for packages that are at 1.0+. I can't remember seeing that documented though (it just fails when you try to register).

[hyrodium]

Okay, I'll update the minor version and remove the deprecated methods in this PR.

[mcabbott]

The logic, I think, is that bumping only the patch version means there are no intermediate versions you can use to later release bugfix supporting older Julia.

I think there is no chance of doing that, and so would vote for 0.5.5. Making a breaking change just for this seems like imposing unnecessary hassle on everyone who depends on it.

On the other hand, now I see how many deprecated methods there are. Cleaning those up is nice. Are there other things waiting which might want to be in the same breaking release?

[hyrodium]

I think #83 will also be considered breaking.

EDIT: The PR was merged just now

[timholy]

#83 is breaking anyway and just got bumped and merged, so I say we just go for it. Should we enforce a 1.6 minimum or are we content to keep supporting 1.0?

[mcabbott]

May as well do 1.6, I think. Even if nothing needs it yet, it'll save bumping again when someone thinks of a feature that wants it.

[hyrodium]

Okay, I'll drop the support for Julia 1.0 👍

[timholy]

We might also consider calling this 1.0?

[mcabbott]

Maybe #75 deserves another look, too?

[timholy]

Oh yeah. I vote we just rip those methods out, they seem really busted.

[mcabbott]

Not fresh in my mind, but maybe #40 (isless vs <, to match AcceleratedArrays.jl) is also worth a look.

[timholy]

We can merge this and then separately decide whether to call the next release 0.6 or 1.0.

[timholy]

Thanks @hyrodium !