You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Seems weird that the behaviour is different on Linux and Windows.
Is it just coincidence that jl_try_substrtod works with the rounding mode set?
(And by the way, is there any particular reason that jl_try_substrtod is written in C instead of Julia?)
Consider the following
On my Windows machine it yields
xdown == xup
, which is unexpected as0.1
can not be represented exactly in binary form.More generally it seems like whatever was told to
setrounding
,parse
does the same rounding, as indicates the following (dirty) test:On Windows it yields
c == 100
(the result of the parsing is always the same) while on Linux it yields the expectedc == 0
.The Linux results were obtained using JuliaBox (for convenience) and the
versioninfo()
on my Windows machine is the following:The text was updated successfully, but these errors were encountered: