A useful feature would be the option to automatically scale some mixing features to add to 1.
Schematically:
(t,g,b) ->(r,g,b)/(r+g+b)
The specific case I have in mind is the channel mixer when converting to B&W: it is redundant, since setting source proportions that add to an arbitrary value are equivalent to setting them to add to 1, then adjusting over-all with exposure. Similar issues would apply for color zones, and for operations in color.
Admittedly, with floating point processing this should make no functional difference, but might avoid the user seeing strange settings where eg the exposure is low but the output is high because of setting all channel values high.
It would also avoid the peculiar situation where the image goes black because all channels start from a default of zero.
Alternatively, a lot of the advantage would come from simply allowing direct entry into the combo box channel values.
Meanwhile, applause for the work currently being done on dt3.0 