In article <20050727.16043418@[EMAIL PROTECTED]
>,
Andreas Lerch <andreas@[EMAIL PROTECTED]
> wrote:
>
>
>>>>>>>>>>>>>>>>>>> Ursprüngliche Nachricht <<<<<<<<<<<<<<<<<<
>
>Am 27.07.05, 11:14:30, schrieb "Kasperer" <john_kasperer@[EMAIL PROTECTED]
>
>zum Thema Playing with STCKCONV, STCKEVAL parameter:
>
>
>> Has somebody already played with STCKCONV, STCKEVAL parameter?
>
>> 1) With a tod value equal to x'FFFFFFFFFFFFFFFF':
>> STCKCONV STCKEVAL=ETODSTMP,CONVVAL=TMEDTE,
>> TIMETYPE=DEC,DATETYPE=MMDDYYYY
>> ETODSTMP DC XL16'00FFFFFFFFFFFFFFFF00000000000000'
>
>
>> Resulting time & date: 23534737 09172042 (09/17/2042)
>
>> 2) With a larger tod value:
>
>> STCKCONV STCKEVAL=ETODSTMP,CONVVAL=TMEDTE,
>> TIMETYPE=DEC,DATETYPE=MMDDYYYY
>> ETODSTMP DC XL16'01000000000000001100000000000000'
>
>> Resulting time & date: 00000000 01011900 (01/01/1900)
>
>> Now, my impression was that STCKCONV with STCKEVAL
>> was especially designed to convert tod values larger
>> than 64 bits. Or did I do something wrong? Thanks.
>
>Hello
>no -- if you take a look of EIBDATE in CICS, you will see the century
>bit. There is a compatibility to IBM-360 systems for yaer 2000
>problem :-)
>yes -- I think, there are a lot of bits to calculate the clock. The
>counter is designed between the cost of bits and the range to
>calculate. :-) would you live since 2053 :-)
>
>one other question: why does several unix-systems does there base at
>01.01.1971 (i think, but it can be a mistake)
Thu, 01 Jan 1970 00:00:00 GMT
> and take a rollover at
>2021 (that can be a mistake too) -
Tue, 19 Jan 2038 03:14:07 GMT
After which, the count of seconds overflows into the sign bit of a 32-bit
value.
There _were_ similar "32-bit" limits in a _lot_ of places inside UNIX.
e.g. a 2gigabyte limit on the size of a single file.
Almost all UNIX variants -- and 'look-alikes' -- today have a
'compatibility
mode' for legacy applications, _with_ the 32-bit limits, *and* a modern
(i.e., '64-bit clean') mode -- where those previously 31/32-bit values
have
been allocated to storage units of 64 bits. "Source-code transparent"
changes, simple re-compilation in the 'new' O/S environment is all that
is required at the code level. Data sets _can_ require "conversion",
however.
> its the cost of bits
>so you can see: it is the count of bits, which can represent the count
>of TOD :-))
>
>bitte nicht ganz so ernst nehmen (sorry, i could not say this sentence
>in english, i am not perfect) my answer is a - good meaning - type of
>comic....
>
>Einen schoenen Tag
>Andreas Lerch
>--
>some days in assembler, some days in date-/time-calculation - since
>1976 :--))
>
>
>
|
