in general i will use expr inside shell scripts for doing arithmetic operations.
is there a way where we can come up with arithmetic operation in a shell script without using expr?
3 Answers
Modern shells (POSIX compliant = modern in my view) support arithmetic operations: + - / * on signed long integer variables +/- 2147483647.
Use awk for double precision, 15 siginificant digits It also does sqrt.
Use bc -l for extended precision up to 20 significant digits.
The syntax (zed_0xff) for shell you already saw:
a=$(( 13 * 2 ))
a=$(( $2 / 2 ))
b=$(( $a - 1 ))
a=(( $a + $b ))
awk does double precision - floating point - arithmetic operations natively. It also has sqrt, cos, sin .... see:
http://people.cs.uu.nl/piet/docs/nawk/nawk_toc.html
bc has some defined functions and extended presision which are available with the -l option:
bc -l
example:
echo 'sqrt(.977)' | bc -l
Comments
Did you tried to read "man ksh" if you're using ksh?
"man bash", for example, has enough information on doing arithmetics with bash.
the command typeset -i can be used to specify that a variable must be treated as an integer, for example typeset -i MYVAR specifies that the variable MYVAR is an integer rather than a string. Following the typeset command, attempts to assign a non integer value to the variable will fail:
$ typeset -i MYVAR
$ MYVAR=56
$ echo $MYVAR
56
$ MYVAR=fred
ksh: fred: bad number
$
To carry out arithmetic operations on variables or within a shell script, use the let command. let evaluates its arguments as simple arithmetic expressions. For example:
$ let ans=$MYVAR+45
echo $ans
101
$
The expression above could also be written as follows:
$ echo $(($MYVAR+45))
101
$
Anything enclosed within $(( and )) is interpreted by the Korn shell as being an arithmetic expression
2 Comments
jim mcnamara
In all fairness, the ksh man page is, um, not so helpful. The ksh88 man pages are even misleading, for example, explaining the _ variable, they tell you it relates to MAILPATH, but do not clearly explain what it is. (on solaris 9, HPUX 11.0 & 11i) It is the argument(s) to the previously entered command, if it matters.
The precision and range of shell builtin $(( <expr> )) depends on the platform. On a 32-bit platform it's limited to 32-bit integers (unlike expr, which appears to support larger integers in a way that appears to be independent of platform).
(N.B - beware that while ^ is a valid operator in most shells, it does not mean 'to the power of', but rather XOR.)
Also - some shells are more forthcoming in this regard. Zsh, for example, allows floating point operations, such as $(( 10/3.0 ))
bcinstead" isn't quite the answer you're looking for...PI=$(echo "4*a(1)" | bc -l)