Re: Random String Generator

[Thorbjørn Ravn Andersen <thunderaxiom@xxxxxxxxx>]

Ralph Daugherty skrev  den 10-09-2008 19:02:
>     yeah, Nathan, I was referring to RPG procedures on that time, I 
> think the timestamp if I recall. But I also recently worked with IFS 
> timestamps (epoch-1970) from RPG API calls and SQL Server timestamps set 
> from Java timestamps (with 9 places provided for resolution of seconds) 
> via RPG JNI, so it's all blurring together by now.
>
>     I do recall it was from an RPG program timestamp function that I'm 
> referring to where time is only populated to milliseconds, and 
> definitely not the gettimeofday() API. I haven't used that yet.
>
>     But regarding the epoch-1970, I was working with the normal seconds 
> since 1970 count stored in file timestamp. The gettimeofday() API must 
> use the extended epoch count that joins in another integer for second 
> precision count.
>
>     I don't have any idea how precisely that's set, but it'd be 
> interesting if RPG time is set to only milliseconds but timeofday() 
> integers set to greater precision than that, like microsecinds. I would 
> think the source would be the same, an RTC chip or whatever. (oh wait, 
> this is the age of the internet - a SOAP call to the National Time 
> whatchmajiggy to get an atomic crystal controlled count, plus or minus a 
> little slack for the SOAP call. ok, plus a lot of slack. :)
>
>   
The "lot of slack"-problem has been solved pre-SOAP in the ntp 
protocol[1], so you need to write a ntp-client and call that instead :)


[1] Feel free to point out the redundancy.