Inconsistent order between FloatingPoints and Rationals

[blkqi]

See the "Note" section of the online documentation for round():
http://julia.readthedocs.org/en/latest/stdlib/math/#Base.round

When trying this on julia v0.3.10 I discovered the following inconsistency which seems to break the trichotomy law.

julia> x = 1.15
1.15

julia> @sprintf "%.20f" x
"1.14999999999999991118"

julia> x < 115//100
false

julia> x > 115//100
false

julia> x == 115//100
false

[tkelman]

Can you try this on an 0.4-dev nightly? I think @simonbyrne fixed this a few months ago on master.

[blkqi]

No problem. This is not an issue using the latest nightly: julia-6eb9133679.

[yuyichao]

Should the fix be backported?

[tkelman]

Would have to dig up which PR it was ("fix comparisons between floating point and rationals" or something like that), but at this point I'd rather not make substantial behavioral changes on release-0.3.

edit: #9198 uses staged functions apparently

[yuyichao]

Wasn't too hard given the author and the keywords. #9198.

Indeed looks a little too big (changes and lines) to be backported for me....