[asterisk-dev] Named lists: app submission

[John Todd]

>On Tuesday 14 August 2007 18:04:38 John Todd wrote:
>>  My company has built app_list.c, which creates named arrays.  It's
>
>Something like that already exists.  It's called HASH(), and it lives in
>trunk, in funcs/func_strings.c.
>
>--
>Tilghman


Ah, I had not understood HASH to work in that way - now it makes a 
bit more sense.  Thanks for the pointer!  I had always seen it as an 
adjunct to ODBC, and not as a standalone string function.

Looking at HASH (and ClearHash, and HASHKEYS) I see how we might have 
been able to use those functions/app to do what we wanted, but I'm 
not disappointed that this app was written.  It seems to be a little 
less effort on the dialplan side.  HASH seems to have the following 
differences from app_list:

1) HASH doesn't have the ability to delete and shift within the list
2) HASH can't push/delete values into/from the list at arbitrary points and
     push the rest of the stack downwards/upwards
3) HASH can't set a bunch of values at once (minor issue)
4) HASH can't report on the length of the non-null values in the list

It seems that our application works more like a stack or heap than 
how HASH works.  The iteration involved in making HASH collapse and 
expand data in a hashname array seems to imply some additional 
dialplan code to make it work like app_list, but I suppose that some 
additional complexity could make it mimic those features.  It does 
seem, however, that HASH might easily be modified to support these 
pop/push features.

Donny:
   Yes, we'll look at the ast_datastore struct.  Some way to prevent 
memory leakage seems to be a Very Good Idea.


PS: Mildly related - does anyone other than me find the function 
"ARRAY" to be confusingly mis-named as far as historical contextual 
meaning of the word "array" in other programming languages?

JT