Re: [Help-bash] Updating global variable from recursing function.

[Frans de Boer]

On 04/02/2012 09:17 PM, Greg Wooledge wrote:
> On Mon, Apr 02, 2012 at 09:08:12PM +0200, Frans de Boer wrote:
> > Strange, it is working fine, the counter is incremented but the result
> > is never propagated to the real global variable, as shown by the bash -x
> > output.
>
> imadev:~$ cat foo
> #!/bin/bash
>
> foo() {
>    local i=7
>    if ((i>  max)); then max=$i; fi
> }
>
> max=0
> foo
> echo "max is $max"
>
> imadev:~$ ./foo
> max is 7
>
>
> > Every time the function returns to the caller, the iHiMDirNest is
> > restored to the level it was before being calling.
>
> That suggests that the variable is in fact local, in some scope or
> other.  Are you sure you showed us the *uppermost* scope?  For example,
> this would do it:
>
> imadev:~$ cat bar
> #!/bin/bash
>
> bar() {
>    local max
>    foo "$@"
> }
>
> foo() {
>    local i=7
>    if ((i>  max)); then max=$i; fi
>    echo "inside foo, max=$max"
> }
>
> max=0
> bar
> echo "at global scope, max is $max"
>
> imadev:~$ ./bar
> inside foo, max=7
> at global scope, max is 0
No I did not show you the whole file (it is about 2184 lines total). But 
I made sure that the variable iHiMDirNest is declared at the global 
level, so before other functions and together with other variables.
The thing is that it goes wrong if mdirtree() is called from within. The 
initial mdirtree() is called from another function, so it will always 
run at the third level anyhow (main->func->func)

I tested it on a directory structure of 7 level deep. The counter did 
well but when finally returning the count decreases with every return 
until it stop at the first instance of mdirtree() where the count was 
correct. So, calling from main-func-func worked fine, but when the last 
function repeated itself one or more times, it went wrong.

I try a simple example myself to look if it is due to the code 
size/complexity or something else.

Frans.