if (_linear) {
v = (control_value - _controllable->lower ()) / (_controllable->upper() - _controllable->lower());
} else {
- v = gain_to_slider_position_with_max (control_value, ARDOUR::Config->get_max_gain());
+ v = gain_to_slider_position_with_max (control_value, _controllable->upper());
}
return v;
double
VolumeController::adjust (double control_delta)
{
- double v = _controllable->get_value ();
- double abs_delta = fabs (control_delta);
-
- /* convert to linear/fractional slider position domain */
- v = gain_to_slider_position_with_max (v, ARDOUR::Config->get_max_gain());
- /* adjust in this domain */
- v += control_delta;
- /* clamp in this domain */
- v = std::max (0.0, std::min (1.0, v));
- /* convert back to gain coefficient domain */
- v = slider_position_to_gain_with_max (v, ARDOUR::Config->get_max_gain());
- /* clamp in this domain */
- v = std::max (_controllable->lower(), std::min (_controllable->upper(), v));
-
- /* now round to some precision in the dB domain */
- v = accurate_coefficient_to_dB (v);
-
- if (abs_delta <= 0.01) {
- v -= fmod (v, 0.05);
+ double v;
+
+ if (!_linear) {
+
+ /* we map back into the linear/fractional slider position,
+ * because this kind of control goes all the way down
+ * to -inf dB, and we want this occur in a reasonable way in
+ * terms of user interaction. if we leave the adjustment in the
+ * gain coefficient domain (or dB domain), the lower end of the
+ * control range (getting close to -inf dB) takes forever.
+ */
+
+ /* convert to linear/fractional slider position domain */
+ v = gain_to_slider_position_with_max (_controllable->get_value (), _controllable->upper());
+ /* increment in this domain */
+ v += control_delta;
+ /* clamp to appropriate range for linear/fractional slider domain */
+ v = std::max (0.0, std::min (1.0, v));
+ /* convert back to gain coefficient domain */
+ v = slider_position_to_gain_with_max (v, _controllable->upper());
+ /* clamp in controller domain */
+ v = std::max (_controllable->lower(), std::min (_controllable->upper(), v));
+ /* convert to dB domain */
+ v = accurate_coefficient_to_dB (v);
+ /* round up/down to nearest 0.1dB */
+ if (control_delta > 0.0) {
+ v = ceil (v * 10.0) / 10.0;
+ } else {
+ v = floor (v * 10.0) / 10.0;
+ }
+ /* and return it */
+ return dB_to_coefficient (v);
} else {
- v -= fmod (v, 0.1);
- }
+ double mult;
+
+ if (control_delta < 0.0) {
+ mult = -1.0;
+ } else {
+ mult = 1.0;
+ }
+
+ if (fabs (control_delta) < 0.05) {
+ control_delta = mult * 0.05;
+ } else {
+ control_delta = mult * 0.1;
+ }
+
+ v = _controllable->get_value();
+
+ if (v == 0.0) {
+ /* if we don't special case this, we can't escape from
+ the -infinity dB black hole.
+ */
+ if (control_delta > 0.0) {
+ v = dB_to_coefficient (-100 + control_delta);
+ }
+ } else {
+ static const double dB_minus_200 = dB_to_coefficient (-200.0);
+ static const double dB_minus_100 = dB_to_coefficient (-100.0);
+ static const double dB_minus_50 = dB_to_coefficient (-50.0);
+ static const double dB_minus_20 = dB_to_coefficient (-20.0);
+
+ if (control_delta < 0 && v < dB_minus_200) {
+ v = 0.0;
+ } else {
+
+ /* non-linear scaling as the dB level gets low
+ so that we can hit -inf and get back out of
+ it appropriately.
+ */
+
+ if (v < dB_minus_100) {
+ control_delta *= 1000.0;
+ } else if (v < dB_minus_50) {
+ control_delta *= 100.0;
+ } else if (v < dB_minus_20) {
+ control_delta *= 10.0;
+ }
+
+ v = accurate_coefficient_to_dB (v);
+ v += control_delta;
+ v = dB_to_coefficient (v);
+ }
+ }
+
+ return std::max (_controllable->lower(), std::min (_controllable->upper(), v));
+ }
- /* and return it */
- return dB_to_coefficient (v);
}