Hi all,
I was curious as to what the experience of others has been in regard to the use of multiple conditioned responses (I'm going to shorten this to MCR for this post).
Personally, I tend to avoid them as I find them too cumbersome to debug and quite honestly, I'm not convinced that they provide much benefit. Perhaps they aren't too bad for two or three responses, but when you have many responses I find you loose the ability to quickly pinpoint the response that is giving an error when trying to debug a MCR. If I instead use a branched dialog tree to evaluate the various conditional responses, I can always know what node if giving me an issue and jump to it from the "Try it out" area. If I try this same thing with MCR I am taken to the main node and then have to manually try to figure out which response is causing the problem.
The other issue that I have with MCR's is that the display in the node is quite frustrating. Firstly, the fact that a node has a whole bunch of conditional responses embedded in it isn't terribly obvious when you are traversing your dialog tree. Secondly, you really only get a small amount of information on the main node screen when you click on a MCR node to view the conditional responses and then you have to go into the settings to see more detail (like what the response does in the "Then assistant should" section. I just find it much clearer to see a branched dialog tree and be able to quickly scroll down the nodes to see what is being evaluated and the full detail on what is happening is only a quick, single click away.
Perhaps MCR's might be needed if your dialog tree gets too big and starts hitting the limits of the system for the number of nodes, but outside that I'm probably going to stick with my conditional branch structures for now.
How about you? I'd love to hear other opinions on the use of MCR's and if you find them useful.
Cheers,
Graham
------------------------------
Graham Walker
------------------------------
#WatsonAssistant