Patch Set 3: (2 comments)
re "you don't mention duplicate avoidance" -- is that asking to improve the commit log? agreed. https://gerrit.osmocom.org/#/c/3979/3/src/vty/command.c File src/vty/command.c: Line 158: if (!*node->name) > use node->name[0]? But why derive it from the prompt? It seems like an inte node->name is a char[], and if it is unset, it starts with 0 as from struct initialization. Hence an unset node->name means !*node->name. Derive it from the prompt: we have a node struct that defines the node ID, a prompt and a write cb, and its commands. The way we enter this node though is not tied to this struct at all. At some other node there is this "arbitrary" vty command that decides: now let's go to this node. My first idea is to try to derive the node name from the command that enters it, but it is impossible to find out. static struct cmd_node msc_node = { MSC_NODE, "%s(config-msc)# ", 1, }; DEFUN(cfg_msc, cfg_msc_cmd, "msc", "Configure MSC options") { vty->node = MSC_NODE; return CMD_SUCCESS; } install_element(CONFIG_NODE, &cfg_msc_cmd); >From msc_node, it is impossible to find out that cfg_msc_cmd which entered the >MSC_NODE is in fact called 'msc'. The next idea is that each cmd_node struct that is created should provide a node name like: static struct cmd_node msc_node = { MSC_NODE, "%s(config-msc)# ", 1, .name = "config-msc", }; but none of your current users of the VTY API do that. We would start off with most nodes being unnamed and it would take some time to find all of them. The nice way would be to define a name, and let the prompt be derived from that. But for hysterical raisins, it is easier to go the other way: all of our nodes have a prompt, and by stripping the special prompt characters we can get a fairly unique name from it without touching any of the dependent code. Line 612: same_name_count = 1; > Why start with 1? For the first node it has not used before? The idea is to get IDs like name name_2 name_3 i.e. the first is just the name, the second gets _2 attached. When I start searching, I know that one 'name' already exists, if I find another one I know that 2 exist. I considered starting with 0, but don't like the result like: name name_1 name_2 nor name_0 name_1 name_2 The _2 suffix is just to make sure we number multiple occurences and should normally not happen. It also is a fallback for empty node names, they will become _1 _2 _3 ... -- To view, visit https://gerrit.osmocom.org/3979 To unsubscribe, visit https://gerrit.osmocom.org/settings Gerrit-MessageType: comment Gerrit-Change-Id: I8fa555570268b231c5e01727c661da92fad265de Gerrit-PatchSet: 3 Gerrit-Project: libosmocore Gerrit-Branch: master Gerrit-Owner: Neels Hofmeyr <[email protected]> Gerrit-Reviewer: Holger Freyther <[email protected]> Gerrit-Reviewer: Jenkins Builder Gerrit-Reviewer: Neels Hofmeyr <[email protected]> Gerrit-HasComments: Yes
