I know what the problem is but there is no real solution. I have found that if I load the payloads one way it can sometimes cause an error with certain complex* commands. If I load the payloads another way (less efficient) it may work with those commands and not with others.
So VC will try the efficient way first, and if that fails, it will try the less efficient way, and if that fails, it should at least tell you which group is causing problems, and then load everything else. I think it only double-fails if you put a bunch of complex* commands into a single group.
I am 99.9% sure that this is a glitch in the windows speech API. They probably don't know about it, probably will never find out, and probably wouldn't care even if they did.
Anyway I feel the problem is basically solved. VC does the most efficient thing that works, and if that fails you just need to break up your commands into groups. Until we encounter a (real) command that won't work even when it is in its own group, I'm not going to spend waste any more time (or hair
) on it.
Note that you can break your group up into multiple groups of the same name. That way you can still use the same enable and disable group commands and it should enable/disable them all. If not there are new commands which enable and disable groups using substring matching.
(* by complex commands I mean commands that use multiple XML pair payloads where phrases are mapped to values.)