I'm trying to re-write a script that we use to monitor some switches in our lab. The number is growing, and I need to start using a loop to check the different stacks.

My problem is that I cannot get the variable name to change and get the value written to the new variable dependent upon the for loop's iteration.

Here is one of the snippets I'm trying to re-script:

As you can see, I'm trying to take the value returned from the snmpget and assign it to a variable whose name is determined by the values of the iteration of the loop.

It should create variables named stack1unit1, stack1unit2, stack1unit3, and stack1unit4 whose values are obtained from the snmpget output.


As you can probably tell, I can't get it to work (snmp stuff works fine-I need no help on that part - it is the variable creation from the loop iteration that I can't get working). I have tried eval with no luck, and can't wrap my head around how indirectly referencing the variable could help me.

Also, if we do manage to create these variables properly, will they still be available once the loop terminates?

Can anyone please shed some light on this for me?

Thanks!

what does the output of $(...) look like?

What have you tried? What did you get? Were there any error messages?

this works for me:
have you considered using arrays instead?

as long as they are not created in a context executed in a subshell, yes. In your case the variables will exist outside the loop

Try

Why not simply use an array??

Hi.

You can create variables stack1unit1 ... stack1unit4 using, e.g.
You can access their values by $stack1unit1 etc, even outside the loop.
For example:
But I can't see how to access their values by index. For example
did not work.

I think you should definitely use bash arrays for that.

Now that's better than read, should work with local/declare too if you want smaller scopes than all children of the script.

For that use ${!} :
But this is all very clunky anyway, arrays are designed for this very task: looping a group of related data.

Thanks for all of the suggestions!

I'm going to try several of them out right now and will post back with the one I wind up using.

Wow! Good thing to know.. Thanks.

I wound up using Juako's read command - it just seemed to be the easiest.

However, I must admit I do not understand why it works. I know the read command will read a line at a time from standard input, but its the "<<<" that's got me fooled (its not east to google those characters

Can you guys explain to me the mechanics behind the "<<<"?

Thanks!

read is no longer read from standard input but rather from the herestring (<<<). You can look here for details.

"<<<" is "send a string to standard input", called "herestring".

But read up from firstfire's post, that's even easier:

for (dynamic I assume) indirect reference the variables:

For the record, $(..) operates as a subshell. There's no way to access any variables set inside it from the outside. You have to arrange the command so that the substitution itself outputs all the values you want, or so that the value you want is produced outside of the subshell.

But as grail mentioned first, and others have repeated, you really, Really, REALLY should be using an array for this, rather than trying to generate dynamic scalar variable names.

Just use stack1unit[${unitNum}]=, and you will have entries of ${stack1unit[0]} ${stack1unit[1]} ${stack1unit[2]}, etc. that you can use later on. If you need the indexes to be text strings instead, you can even use an associative array (assuming bash 4+)

http://mywiki.wooledge.org/BashGuide/Arrays
http://mywiki.wooledge.org/BashFAQ/006

The second link is for reference only. You really should be using an array for this.


Oh, and by the way, did I mention that you really should be using an array for this?

Yup agreed, (dynamic) indirect vars definitely have their uses where they're better than arrays, but this doesn't seem to be the case IMO.