fetchListFTP: implemented, but crippled

[Dag-Erling Smorgrav]

Tim Kientzle <kientzle at acm.org> writes:
> Easiest fix: move 'name' field to end (which
>     permits variably-sizing it) and redefine API
>     to return a linked-list

That would work.

> Better fix: redesign API to return one
>     entry at a time.  Then client code can manage
>     storage as appropriate.  With this approach,
>     struct url_ent could remain unchanged.

You'd run into trouble if the caller tried to perform another
operation before reaching the end of the list.

> The current API makes fetchListFTP pretty much
> unusable for me; I'm not comfortable with that
> kind of memory usage.  If there's any concensus
> about which fix is appropriate, I can start
> putting it together.

Have you tried contacting the author?

> +	/* Use last whitespace-separated token as filename */

That won't work for symlinks.

> +		if(url_index >= url_list_length) {
> +			new_url_list_length = url_list_length * 2;
> +			new_url_list = malloc(sizeof(struct url_ent) 
> +					      * new_url_list_length);
> +			fprintf(stderr,"fetchListFTP: malloc(%d)\n",
> +				sizeof(struct url_ent) 
> +				* new_url_list_length);
> +			memcpy(new_url_list, url_list,
> +			       sizeof(struct url_ent)*url_list_length);
> +			memset(new_url_list+url_list_length,0,
> +			       sizeof(struct url_ent)
> +			       *(new_url_list_length - url_list_length));
> +			free(url_list);
> +			url_list = new_url_list;
> +			url_list_length = new_url_list_length;
> +		}

You know, we have realloc() for a reason...

DES
-- 
Dag-Erling Smorgrav - des at ofug.org