Re: Motion P1788.1/M004.01
On 17 May 2016, at 18:42, Michel Hack <mhack@xxxxxxx> wrote:
> On Tue, 17 May 2016 19:22:10 +0000, John Pryce wrote, replying to me:
>>> What I meant was that none of the required functions in 1788.1 produce
>>> semibounded interval results -- except, of course, constructors.
>>
>> The intersection and convexHull operations of 6.7.3 can do. And at
>> Level 2, so can arithmetic operations if overflow occurs at one end.
>
> The context was (in light of the issues raised by Lee Winter) TRULY
> unbounded, as opposed to the result of overflow (which is still bounded,
> even if the bound cannot be determined). Intersection and ConvexHull
> cannot create semibounded intervals from bounded intervals.
Hang on: doesn't the reciprocal of [0,1], recip([0,1]), equal [1,oo] according to the 1788 definition?
John P