|
View previous topic :: View next topic
|
| Author |
Message |
truly198
New User
Joined: 15 May 2010
Posts: 4
Location: CN
|
|
|
|
as captioned, Ie. S9(9) COMP.
Is this correct if i deduce this way -
it takes up 4 bytes. so 4 * 8 =32 bits
so the maxmum is 2 32 - 1 = 4294967295
or the sign takes one bit 2 31 - 1 = 2147483647 |
|
| Back to top |
|
 |
Akatsukami
Global Moderator
Joined: 03 Oct 2009
Posts: 1787
Location: Bloomington, IL
|
|
|
|
| Yes, that is correct. |
|
| Back to top |
|
|
Bill Woodger
Moderator Emeritus
Joined: 09 Mar 2011
Posts: 7309
Location: Inside the Matrix
|
|
|
|
Yes, four bytes. A Word, or Full-Word, of storage.
The sign is the high-order bit. Everything above 7FFFFFFF is -ve. I'm not going to check your maths, it looks about right.
So, with Cobol, the point is that for the size of a binary field, the compiler will go to the smallest out of half-/full-/double-word. If you go for S9(10) you will get a double-word of storage allocated. If you go for S9(4) a half-world, S9(5) a full-word, like your example.
S9 to S9(4), half-word
S9(5) to S9(9), full-word
S9(10)+, double-word
Turns out, S9(9) is about the worst possible thing to use for binary calculations. |
|
| Back to top |
|
|
Robert Sample
Global Moderator
Joined: 06 Jun 2008
Posts: 8700
Location: Dubuque, Iowa, USA
|
|
|
|
| Quote: |
as captioned, Ie. S9(9) COMP.
Is this correct if i deduce this way -
it takes up 4 bytes. so 4 * 8 =32 bits
so the maxmum is 2 32 - 1 = 4294967295
or the sign takes one bit 2 31 - 1 = 2147483647
|
Maybe. COBOL has the TRUNC option that can be set differently at each site. Depending upon the TRUNC option, a PIC S9(9) COMP variable may have the range -2,147,483,648 to 2,147,483,647 OR -999,999,999 to 999,999,999.
Without looking at the compile options, you cannot tell which it is -- and since the TRUNC option can be changed for each compile, you cannot state an absolute rule for the variable range -- it will depend upon the TRUNC setting when the compile is done. |
|
| Back to top |
|
|
Bill O'Boyle
CICS Moderator
Joined: 14 Jan 2008
Posts: 2501
Location: Atlanta, Georgia, USA
|
|
|
|
If your COBOL compiler supports COMP-5 (OS/390 COBOL 2.2.1 and greater), then the TRUNC option doesn't matter as COMP-5 (Native Binary) ignores the TRUNC option.
However, as Robert has said, the TRUNC option can have a surprising effect on the maximum-range in binary variables.
The TRUNC option to use for maximum-range (for non COMP-5), is TRUNC(BIN).
Bill |
|
| Back to top |
|
|
Bill Woodger
Moderator Emeritus
Joined: 09 Mar 2011
Posts: 7309
Location: Inside the Matrix
|
|
|
|
| Good point Robert. Effectively TRUNC is telling the compiler where to use the space allocated to determine the maximum value (your example) or the PICture to determine the maximum value, +/-999,999,999. Check out COMP-5 as well in the manual. |
|
| Back to top |
|
|
dbzTHEdinosauer
Global Moderator
Joined: 20 Oct 2006
Posts: 6965
Location: porcelain throne
|
|
|
|
keep in mind, (cobol 2, trunc(bin))
when you move a literal 1,000,000,000 to a comp field,
the comp field must be defined as s9(10) or 9(10), which is a double word.
you can then move the doubleword to a single word comp field, [s9(9) or 9(9) comp]
and the trunc(bin) option will perserve the integrity of the value.
in other words, your literal and the receiving field have to match. |
|
| Back to top |
|
|
Bill Woodger
Moderator Emeritus
Joined: 09 Mar 2011
Posts: 7309
Location: Inside the Matrix
|
|
|
|
| Bill Woodger wrote: |
| Good point Robert. Effectively TRUNC is telling the compiler where to use the space allocated to determine the maximum value (your example) or the PICture to determine the maximum value, +/-999,999,999. Check out COMP-5 as well in the manual. |
Silly fingers.
Good point Robert. Effectively TRUNC is telling the compiler whether to use the space allocated to determine the maximum value (your example) or the PICture to determine the maximum value, +/-999,999,999. Check out COMP-5 as well in the manual.
In addition, it doesn't happen by magic that you get different maximum values. If you use the compile option to set the PICture as definining the maximum value, then the compiler will generate code, at each point at which the value of the field can change, to ensure that the value is not become greater than the PICture size (ie high-order tuncation).
Along with Dick's point about literal values (and try geting 1,000,000,000 on a VALUE clause with either compile option) also watch out for "edited" fileds for printing/display. Although you might have S9(9) as your PICture, you would need Z(9)9 (or equivalent, 10 printable digits) to ensure displaying everything with NOTRUNC.
While we are at it, again, the manual suggests avoiding binary fields with a PICture of 9 digits in situations where performance is important. For calculations, this is first converted to a double-word, then uses double-word artihmetic, then converted back to a full-word. A PICture with 10 digits is already a double-word, so avoids the two conversions. A PICture with 8 digits will use full-word arithmetic. |
|
| Back to top |
|
|
Bill Woodger
Moderator Emeritus
Joined: 09 Mar 2011
Posts: 7309
Location: Inside the Matrix
|
|
|
|
Off-topic, but, just for fun, can you tell the range at run time?
| Code: |
move 1 to value
loop until
( ( value is zero )
Or
( value is negative ) )
keep previous value
add 1 to value
end-loop
previous value is positive limit
if value is negative
value is negative limit
else
if previous value / 2 is exact (and gets here)
value is positive only
else
test value = 0 - previous value
if previous value = test value
value is positive only (with all 9's)
else
test value is negative limit
end-if
end-of
end-if
|
If it works and you don't know why, someone can explain. |
|
| Back to top |
|
|
Bill Woodger
Moderator Emeritus
Joined: 09 Mar 2011
Posts: 7309
Location: Inside the Matrix
|
|
|
|
| Bill Woodger wrote: |
Off-topic, but, just for fun, can you tell the range at run time?
[...]
|
An error, a why-do-it-like-that and a typo.
| Code: |
"value" is a COMP field of any size and any picture.
The idea is, using only value, to establish the positive/negative limits at run-time.
move 1 to value
loop until
( ( value is zero )
Or
( value is negative ) )
add 1 to value
end-loop
subtract 1 from value
display value as positive limit
add 1 to value
if value is negative
display value as negative limit
else
value = ( 0 - ( value - 1 ) )
if value is positive
* value is positive only
else
display value as negative limit
end-if
end-if
|
|
|
| Back to top |
|
|
|
|
  |
All times are GMT + 6 Hours |
|
