On Sat, Oct 6, 2018 at 1:23 PM Michael Osipov <[email protected]> wrote:

> Am 2018-10-06 um 21:15 schrieb Gary Gregory:
> > On Sat, Oct 6, 2018 at 12:57 PM Michael Osipov <[email protected]>
> wrote:
> >
> >> Am 2018-10-06 um 19:37 schrieb Oleg Kalnichevski:
> >>> On Sat, 2018-10-06 at 10:03 +0200, Michael Osipov wrote:
> >>>> I like it a lot too, but why are internal value retained as int/long
> >>>> and
> >>>> not use the TimeValue clasS?
> >>>>
> >>>
> >>> Should they be?
> >>
> >>   From a consistency point of view doesn't it make sense? Otherwise if
> >> the int/long is an implementation detail and is not exposed outside that
> >> is perfectly fine I think.
> >>
> >
> > One thing that sucks is that TimeUnit millis are longs and that low level
> > APIs like Socket timeout take int millis.
>
> Yet, that is awkward throughout the entire JDK. I'd expect long
> everywhere where milliseconds have to be provided. E.g.,
> Thread#sleep(long).
>

FWIW, I'm all for using TimeUnit in all APIs, not just the public ones.
Then in whatever crucial spots we can deal with int vs. long.

Gary

>
> M
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: [email protected]
> For additional commands, e-mail: [email protected]
>
>

Reply via email to