On Fri, 21 Mar 2008 18:38:19 -0400 marius schebella marius.schebella@gmail.com wrote:
want to delay by 1000/44100 milliseconds and not 1/44100 milliseconds.
Ah yes, sorry that's a typo as I rushed those examples together.
and then, there is this problem that delwrite~/delread~ is sensitive to the order of creation. I attached a small patch, you can see that a delwrite/read that is created before the line~ object works similar to z~, wheras the delwrite/delread that is created after the line~ is one vector behind.
This is much more like it, good thinking Marius. One block of difference would explain the magnitude discrepency. And that's the sort of thing I really want to avoid having to explain in the text.
Claude also pointed out a possible sign discrepency around zero for [delta~] which uses IEEE floats instead of t_sample. This one is most dangerous if the object is followed by a [wrap~]
Clearly these methods are subtley different and if you're using them in a sensitive context the effects are important.
I'm going to stick with your first suggestion of [fexpr~], because it works well, and you've made me realise it's a useful object to introduce.
cheers,
Andy