Closed lbenet closed 6 years ago
Thanks, LGTM. Not sure what's up with the bisect
function.
The problem is not bisect
itself, but using for that test the a..b
notation, which uses prevfloat(a)
if a
is not an integer and something similar for b
with nextfloat(b)
.
That's the reason we get:
julia> using IntervalArithmetic
julia> 0.5..0.5
[0.499999, 0.500001]
julia> inf(ans), sup(ans)
(0.49999999999999994, 0.5000000000000001)
bisect
(correctly) returns a tuple of intervals constructed with Interval
.
I just submitted a new commit which fixes the test.
Tests pass now in Julia 0.6, but fail in Julia 0.7.
Any ideas?
No, sorry, it's not clear what the problem is.
Can we use FemtoCleaner to get rid of all the deprecations?
We should certainly use it. Perhaps #14 and #17 should be merged as well.
Restarted failing travis jobs on nightly. Let's see if https://github.com/JuliaIntervals/IntervalArithmetic.jl/issues/93 is related...
Closing, since it is currently reexported
Tests are failing after cloning because
Interval
is not being recognized. I simply solve this by adding it as aconst