On Mon, Mar 08, 2004 at 06:30:01PM +0100, Tim Goetze wrote: > we're already requiring hosts to police connections for value > conversions.
Can you point me to some document that contains that requirement ? > >The two ends of a trigger connection are not symmetrical. > > aren't they? can you expand on this, please? Never mind. This would lead us very far. > MOMENTARY: a push button, sending 'yes' when pressed and 'no' when > released. 'when' or 'while' ? > TRIGGER: a push button, sending 'yes' only at press *or* release, and > 'no' at any other point in time. Why would you ever need both ? Unless the two events are one sample apart (unlikely if they come from a GUI), you can always use the first. And if something like "TRIGGER" is used, it should at least send different events for press and release. And in that case "MOMENTARY" is a subset and consequently not required. I'm more and more alarmed by the way this debate is going. I see lots of ideas but no clearly defined concepts nor any attention to completeness and consistency. I see implementation (diffs on ladspa.h) before specification. If such as thing as a new LADSPA version (with feautures that can *not* be ignored by a host without impairing functionality of either the host or the plugin) is introduced, then we should at least sit down, think, and come out with something that is structured and consistent. I'm not saying such a move is necessary. Note that the two changes I proposed - HINT_ENUMERATED, always together with HINT_INTEGER, and appending the labels to an existing array, - HINT_CAN_BE_NULL, to signal that a port pointer can be NULL when unconnected, can be safely ignored by any host. In the fist case, the GUI will show ints instead of labels. In the second, the plugin may waste some CPU cycles. -- FA
