Andy Farnell a écrit :

Doesn't quite work. If you try it with 1/sr in place of [z~] in a patch that's sensitive to it the results are wrong . Not sure why.

this is strange. look at my exemple, it show that it should work. (1 signal is delayed with z~ and with delread/write; diference of both is pure silence)

Could you send the patch you tested?

cyrille

andy

On Fri, 21 Mar 2008 09:55:27 +0100 cyrille henry cyrille.henry@la-kitchen.fr wrote:

...

hello, i think it's quite simple to emulate z~ with a simple delread~ / delwrite~.

cyrille

Andy Farnell a écrit :

...

I have a problem, that is partly of my own making and need to air this on the list for advice. This isn't a rant, just a frustrating situation.

As some of you know I have been working on a textbook for Pd for some time. In the final stages and adding the practical exercises and the problem is this:

The book is completely for vanilla Pd to keep things simple and I really don't want to change this.

I have three crucial exercises that use [delta~]

[delta~] does not exist in vanilla Pd

[delta~] must be recreated _EXACTLY_ as the external behaves.

It IS possible to recreate [delta~] exactly using [z~]

[z~] does not exist in vanilla

It is not, as far as I can see, possible to emulate [delta~] using [rzero~] or any other signal domain method, in a way that doesn't confuse the hell out of the reader or provide unsatisfactory results.

I am left with these options:

1. Drop three of the most interesting exercises from the textbook.

2. Change my policy and make the textbook require pd-extended on account of

only _one_ object!

3. Find a vanilla replacement for [delta~] that is simple. (less than three objects)

4. Convince Miller to include [delta~] in vanilla

5. Convince Miller to include [z~] in vanilla

To the list: I'm hoping some bright spark will solve (3). Be warned, it isn't as simple as it seems, AFAICS. [rzero~] should provide a differentiator response in theory. In practice dealing with the offsets and scales ruins the whole show and leads to over-complex and unstable patches. Using [biquad~] is extremely confusing to most readers and I don't want to do this. You cannot obtain a single sample delay using [delwrite~] as far as I know, if you can it certainly means [samplerate~] must be added to all patches to make sure they run at different sr, which is mucho annoying.

To Miller: Please consider admitting [z~] or [delta~] (I actually prefer the former because it is more educational to see differentiator built from sample delay) into the vanilla distro. This would need to be time for release quite soon. I'm sure it would greatly benefit Pd to do this.

Best regards 2 all,

Andy