The expression to be evaluated. This object is the same
as a DisplayString (RFC 1903) except for its maximum length.
Except for the variable names the expression is in ANSI C
syntax. Only the subset of ANSI C operators and functions
listed here is allowed.
Variables are expressed as a dollar sign ('$') and an
integer that corresponds to an expObjectIndex. An
example of a valid expression is:
($1-$5)*100
Expressions must not be recursive, that is although an expression
may use the results of another expression, it must not contain
any variable that is directly or indirectly a result of its own
evaluation. The managed system must check for recursive
expressions.
The only allowed operators are:
( )
- (unary)
+ - * / %
& | ^ << >> ~
! && || == != > >= < <=
Note the parentheses are included for parenthesizing the
expression, not for casting data types.
The only constant types defined are:
int (32-bit signed)
long (64-bit signed)
unsigned int
unsigned long
hexadecimal
character
string
oid
The default type for a positive integer is int unless it is too
large in which case it is long.
All but oid are as defined for ANSI C. Note that a
hexadecimal constant may end up as a scalar or an array of
8-bit integers. A string constant is enclosed in double
quotes and may contain back-slashed individual characters
as in ANSI C.
An oid constant comprises 32-bit, unsigned integers and at
least one period, for example:
0.
.0
1.3.6.1
No additional leading or trailing subidentifiers are automatically
added to an OID constant. The constant is taken as expressed.
Integer-typed objects are treated as 32- or 64-bit, signed
or unsigned integers, as appropriate. The results of
mixing them are as for ANSI C, including the type of the
result. Note that a 32-bit value is thus promoted to 64 bits
only in an operation with a 64-bit value. There is no
provision for larger values to handle overflow.
Relative to SNMP data types, a resulting value becomes
unsigned when calculating it uses any unsigned value,
including a counter. To force the final value to be of
data type counter the expression must explicitly use the
counter32() or counter64() function (defined below).
OCTET STRINGS and OBJECT IDENTIFIERs are treated as
one-dimensioned arrays of unsigned 8-bit integers and
unsigned 32-bit integers, respectively.
IpAddresses are treated as 32-bit, unsigned integers in
network byte order, that is, the hex version of 255.0.0.0 is
0xff000000.
Conditional expressions result in a 32-bit, unsigned integer
of value 0 for false or 1 for true. When an arbitrary value
is used as a boolean 0 is false and non-zero is true.
Rules for the resulting data type from an operation, based on
the operator:
For << and >> the result is the same as the left hand operand.
For &&, ||, ==, !=, <, <=, >, and >= the result is always
Unsigned32.
For unary - the result is always Integer32.
For +, -, *, /, %, &, |, and ^ the result is promoted according
to the following rules, in order from most to least preferred:
If left hand and right hand operands are the same type,
use that.
If either side is Counter64, use that.
If either side is IpAddress, use that.
If either side is TimeTicks, use that.
If either side is Counter32, use that.
Otherwise use Unsigned32.
The following rules say what operators apply with what data
types. Any combination not explicitly defined does not work.
For all operators any of the following can be the left hand or
right hand operand: Integer32, Counter32, Unsigned32, Counter64.
The operators +, -, *, /, %, <, <=, >, and >= work with
TimeTicks.
The operators &, |, and ^ work with IpAddress.
The operators << and >> work with IpAddress but only as the
left hand operand.
The + operator performs a concatenation of two OCTET STRINGs or
two OBJECT IDENTIFIERs.
The operators &, | perform bitwise operations on OCTET STRINGs.
If the OCTET STRING happens to be a DisplayString the results
may be meaningless, but the agent system does not check this as
some such systems do not have this information.
The operators << and >> perform bitwise operations on OCTET
STRINGs appearing as the left hand operand.
The only functions defined are:
counter32
counter64
arraySection
stringBegins
stringEnds
stringContains
oidBegins
oidEnds
oidContains
average
maximum
minimum
sum
exists
The following function definitions indicate their parameters by
naming the data type of the parameter in the parameter's position
in the parameter list. The parameter must be of the type indicated
and generally may be a constant, a MIB object, a function, or an
expression.
counter32(integer) - wrapped around an integer value counter32
forces Counter32 as a data type.
counter64(integer) - similar to counter32 except that the
resulting data type is 'counter64'.
arraySection(array, integer, integer) - selects a piece of an
array (i.e. part of an OCTET STRING or OBJECT IDENTIFIER). The
integer arguments are in the range 0 to 4,294,967,295. The
first is an initial array index (one-dimensioned) and the second
is an ending array index. A value of 0 indicates first or last
element, respectively. If the first element is larger than the
array length the result is 0 length. If the second integer is
less than or equal to the first, the result is 0 length. If the
second is larger than the array length it indicates last
element.
stringBegins/Ends/Contains(octetString, octetString) - looks for
the second string (which can be a string constant) in the first
and returns the one-dimensioned arrayindex where the match began.
A return value of 0 indicates no match (i.e. boolean false).
oidBegins/Ends/Contains(oid, oid) - looks for the second OID
(which can be an OID constant) in the first and returns the
the one-dimensioned index where the match began. A return value
of 0 indicates no match (i.e. boolean false).
average/maximum/minimum(integer) - calculates the average,
minimum, or maximum value of the integer valued object over
multiple sample times. If the object disappears for any
sample period, the accumulation and the resulting value object
cease to exist until the object reappears at which point the
calculation starts over.
sum(integerObject*) - sums all available values of the
wildcarded integer object, resulting in an integer scalar. Must
be used with caution as it wraps on overflow with no
notification.
exists(anyTypeObject) - verifies the object instance exists. A
return value of 0 indicates NoSuchInstance (i.e. boolean
false). |