KR> # HG changeset patch KR> # User Kaitlin Rupert <karupert@us.ibm.com> KR> # Date 1195082890 28800 KR> # Node ID b4f0488c0d16a228a3c6fd6f23f7465cd3944123 KR> # Parent 6dc0a1a05c60ad08738c7c48aa9fb6d7dd7cd8be KR> Add function to parse InstanceIDs. Great! KR> +bool parse_id(const CMPIObjectPath *ref, KR> + char **name) KR> +{ KR> + char *id = NULL; KR> + char *pfx = NULL; KR> + int ret; KR> + KR> + id = cu_get_str_path(ref, "InstanceID"); KR> + if (id == NULL) KR> + return false; KR> + KR> + ret = sscanf(id, "%a[^:]:%as", &pfx, name); KR> + KR> + free(id); KR> + free(pfx); KR> + KR> + if ((ret != 2) || (*name == NULL)) KR> + return false; KR> + KR> + return true; KR> +} What do you think about making parse_id() take a string id, and then also exposing a wrapper function called parse_instanceid() that takes the ref? I'd hate to bundle the actual parsing into a function that takes such a complex type. Also, since this is a library function and it could potentially be used in other places down the road, I think it would be better to make it return both sides of the id, as parse_fq_devid() does now. Yes, it will require the caller to free() another string, but I think its better that way. -- Dan Smith IBM Linux Technology Center Open Hypervisor Team email: danms@us.ibm.com