[getdns-api] timeouts specified as milliseconds rather than seconds

[Wiley, Glen]

Bear in mind that if we offer ms then one can still have a multi-second
timeout.  If we offer only seconds then the user has no option to use ms
timeout.
-- 
Glen Wiley
KK4SFV

Sr. Engineer
The Hive, Verisign, Inc.




On 11/7/13 3:27 PM, "Melinda Shore" <melinda.shore at nomountain.net> wrote:

>On 11/07/2013 12:12 PM, Matt Miller wrote:
>> I think millisecond precision is actually burdensome to API users in
>> the wild.
>
>I think it really depends on who those API users actually are.  There
>are some tradeoffs here and I think it's valuable to consider several
>things.  I think there's pretty clearly value to some kinds of
>applications in having subsecond granularity.  For example, in VoIP
>address resolution is one of a number of things that contribute to
>delay between the time the user has finished dialing (or whatever) and
>the time that the phone starts ringing.  There are other situations
>in which there's either delay sensitivity or user experience
>issues.
>
>Now, I'd agree that millisecond resolution is almost certainly
>unnecessarily precise, but it's something that Unix/POSIX developers
>are used to working with and it seems reasonable to me that if there's
>interest in subsecond resolution, timeval is a reasonable encoding.
>But ultimately it's about the tradeoffs, and if there's an expectation
>that applications using getdns will simply not need that kind of
>granularity, then it's probably not worth the effort.
>
>Melinda
>_______________________________________________
>getdns-api mailing list
>getdns-api at vpnc.org