[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [sc-dev] [approve] LinLin degenerate condition

On Tue, Feb 21, 2006 at 10:31:21PM -0500, James Harkins wrote:
> OK to commit? If the source and target ranges are the same, LinLin is 
> superfluous. Would this also apply to ExpExp? (Obviously not LinExp.)

and maybe Line.kr(0, 0, doneAction: 0)? just kidding.

i'm not convinced that this sort of micro-optimization of
the graph should be implicitly provided by ugen
constructors; obviously we could go through heaps of ugens
and apply special conditions. IMO it makes the code less
readable and might give surprising results e.g. when tracing
a synth (where's that LinLin?). you also have to carefully
match semantics in the SC class and the ugen code.

if you're writing synthdefs by hand, why not just leave the
LinLin out? if you're generating synthdefs, why not provide
a hook in the synthdef builder, and do any graph
optimizations explicitly?