Jan Schampera wrote: > Well, I guess intuitively expected that it evaluates to 0, like for > example: > > $ test=string > $ echo $((test)) > 0
In that case it is like the 'atoi(3)' routine. A non-numeric has no number value and is evaluated to be zero. A single zero is a valid arithmetic expression. $ echo $((0)) 0 But two numbers is a different case. $ echo $((0 0)) sh: line 1: 0 0: syntax error in expression (error token is "0") This is the same in bash as in other languages too. They are all very similar on this point. > But I just made some experiments (which I should have done before) and > read the manual again a bit closer: In this example I really reference > the variable "string" (as designated operand) which is unset/null and > thus evaluates to 0. > > I also saw this can be done in any depth. Yes. Well, all's well that ends well. :-) > So, excuse my wrong interpretion of the manual (maybe that needs to be > marked extra by a sentence or a half in manual). It's definitely not a > bug (it is - between screen and chair!). Suggestions for improvements to the documentation are always appreciated. > Best regards and thanks again for quick reply, Good luck! Bob _______________________________________________ Bug-bash mailing list Bug-bash@gnu.org http://lists.gnu.org/mailman/listinfo/bug-bash
