Anthony Minessale II added a comment - 01/Mar/12 1:08 PM Did you try it =D The paramater is sent as "Menu-Name" already..... It would have been easy to add but now I don't get the bounty cos it's already there. (since 2009 by Matthieu Rene)

abaci added a comment - 01/Mar/12 2:24 PM Thanks for the info about Menu-Name, however the Bounty is still here as I would still have to load all the ivrs since menu_sub will not make a new request, which I need in my setup. what I would like is a new "ivr" binding which should only be called when requesting an ivr (not just the general "configuration" binding), and when menu-sub is called it should do a new request for that menu (with Menu-Name set to the name of the sub-menu), menu-back should not have to do a new request since we already have it. by doing it as a separate binding we can preserve the old way if someone has a binding for configuration. Let me know if you're interested and I will Paypal you the money. Thanks

E. Schmid added a comment - 01/Mar/12 3:50 PM Creating an new "ivr" binding might be a large request for $50 Would it be easy to have FreeSWITCH send a new XML_CURL request on a menu-sub action?

Anthony Minessale II added a comment - 01/Mar/12 5:23 PM Creating a new request on a sub menu action is much better, the ivr binding idea violates the scope of the xml and config too much. Even doing sub menu reqs is about a $300 bounty.

abaci added a comment - 04/Mar/12 10:38 AM I may be able to go up to $100 to enable separate request for menu-sub. if anyone is interested please let me know asap.

abaci added a comment - 12/Mar/12 1:35 PM anyone interested in implementing this before the 1.2 feature freeze?

Ken Rice added a comment - 30/Mar/12 11:48 AM Reclassed as a feature request.

Cal Leeming added a comment - 01/Apr/13 8:40 AM abaci: Could you explain a little bit about what you are trying to achieve and why? I understand that you want a separate IVR binding, but I do not understand the thought process behind it. When mod_xml_curl requests the IVR configuration, you return back all the relevant menus.. for example if you have 1 root menu and 50 sub menus that it could follow, then you would return all 50. You don't have to return every single IVR menu possible in each request, just the root menu and the relevant sub menus that it can follow. This is easily done if you have a proper ORM framework and database schema.. i.e. you would create an IVR object, then an IVRAction object, then link other IVR menus via an IVRAction object, then push back the menus where IVR.id = X and IVRAction.to_menu_id = X.

abaci added a comment - 03/Apr/13 6:07 PM I know I can return all relevant menus but I wanted to not have to return them all (sometimes it may be more than a few sub menus and each may have it's own set of sub menus) For now I just don't use the menu-sub feature.

Cal Leeming added a comment - 03/Apr/13 6:23 PM May I ask the reasoning behind why you don't wish to return all the relevant sub menus in each request? Also - by not using the menu sub feature, you would lose the ability to jump between menus in breadcrumb logic (i.e. prev/first menu options etc)

abaci added a comment - 03/Apr/13 6:27 PM the reason is to avoid latency of processing all ivr entries generating the xml and then freeswitch parsing the xml. I know I'm losing the ability to jump between menus, that was the point of opening this ticket.

Cal Leeming added a comment - 03/Apr/13 6:35 PM - edited If this request is purely based on performance, have you done profiling to measure the amount of latency difference between returning 1 menu and 100 menus? If so, please put the results in the ticket as well for future reference. If not, then you should do that first, otherwise this request might be unnecessary. Have you considered that having 100 individual HTTP requests rather than 1 HTTP request, could actually be worse? If you add up all the time it takes both server and client side, that's a lot of context switching. This is, of course, all speculation and is further reason why performance tests should be done first. If the request is based on API cleanness, it could also be argued that sub menus should make an individual XML call to make the API feel better (imho). In a nut shell, I think it would be worth doing some tests before hand and get some real numbers, rather than spending money on something which may not be needed at all.

abaci added a comment - 07/Apr/13 12:47 PM I did do some test back when I opened this issue but can't find the results now but loading all menus was definitely more load (in my case at least) and your argument that 100 http requests would be more load is not correct as you would only return the menus that caller navigates which will almost never be all, and even if it would add more load in total it would still improve on latency. I'm in no way saying that this is true for everyone, just reason for me offering a bounty. if no one has an interest in this it can be closed.

Cal Leeming added a comment - 10/May/13 5:49 PM Abaci: it sounds like your request will be at least over $300, and it is not agreed whether this is even the correct way to be doing things. If you post back your maximum bounty offer, we'll see if someone will take it up.