I asked for something like Antoine's design back in 2007. I think it's a great idea, because it behaves like a signal inlet in compiled objects. On Feb 14, 2016 6:48 PM, "Ivica Bukvic" ico@vt.edu wrote:
On Sun, Feb 14, 2016 at 5:54 PM, Jonathan Wilkes via Pd-list < pd-list@lists.iem.at> wrote:
Hi Antoine, We're talking about two different kinds of "dynamic" nlets. Yours seems to set a value for the signal based on whether or not there is a connection to that inlet. What I'm talking about is a future object that would elegantly handle the creation of a variable number of inlets(~) or outlets(~) inside an abstraction.
I'm pressing Ivica specifically on variable signal nlets because there's no way to flexibly handle them.
Why not? First of all, we need to agree that such an object would be mostly useful at instantiation time (e.g. a silly example would be an abstraction that mimics selector~ behavior) and in dynamic patching where after creation one would need to connect bunch of stuff to this object and ensure that inside the abstraction additional inlets actually do something. Second, let's assume the dynamic nlet creation object is the rightmost object (otherwise user is asking for potential problems if something is already connected to the second inlet and then suddenly first inlet grows a couple more inlets before the second one--even this could be potentially handled, with adequate changes inside the pd core, or in this case pd-l2ork core). Now, if new nlets are generated, they will not autoconnect to anything--this is user's responsibility likely through some sort of live or scripted patching at instantiation time which could be driven from the same argument. Once that is all taken into an account, the last thing is to consider:
suspend dsp instantiate new object rebuild dsp graph (as you would with instantiation of any new signal object resume dsp redraw object and parent object on its parent canvases
One lingering concern is that this would effectively make the abstraction dirty which could be either seen as a good thing or handled as something that does not trigger the dirty flag.
Best,
Ico
-Jonathan
On Sunday, February 14, 2016 5:34 PM, Antoine Rousseau < antoine@metalu.net> wrote:
I've only partially followed all this discussion (not using Max myself), but maybe an object I wrote could help you building such abstractions :
[moonlib/dinlet~] is an [inlet~] with an init float value (constant signal) as an argument. This default value is overloaded when a signal is connected to the inlet, but restored when the signal is disconnected. A float sent to it would overwrite the default constant value.
Of course the init default value could be one of the abstraction's arguments ($xxx)...
BUT :
- there is a very little hack (which could be called a bugfix...) that
has to be made to pd source (this change is written in comment in the source file of dinlet~). I should open a ticket for that in the sourceforge repo. The involved bug is mixing the different float values up when [dinlet~] is used together with normal [inlet]s.
- I should add a missing feature in dinlet~, which would add an inlet to
the [dinlet~] object itself, to allow changing the default value inside of the abstraction.
If anyone think this would be helpful, I could do this (open a ticket and update moonlib about this missing inlet).
2016-02-14 20:29 GMT+01:00 Jonathan Wilkes via Pd-list < pd-list@lists.iem.at>:
Why not simply have an inlet that can handle both inside an abstraction
and route signal one way and number the other and then sprinkle that with dynamic nlet creation and you're done? Then you can simply abstract most cases.
I read (and like) your spec on dynamic nlet creation, but I have a problem with section 2.1 Signals:
"To handle the dynamic creation of signal inlets and their routing within the abstraction, the implementation must"
It looks like the rest of the section is missing. :)
-Jonathan
On Sunday, February 14, 2016 1:51 PM, Matt Barber brbrofsvl@gmail.com wrote:
I tried coding that once, but it seemed like it needed some big change in architecture. Technically it's only the main signal that accepts both messages and signals in this way, where you would want to route the message. Floats should almost always be promoted to signals.
On Sun, Feb 14, 2016 at 1:18 PM, Ivica Ico Bukvic ico@vt.edu wrote:
Why not simply have an inlet that can handle both inside an abstraction and route signal one way and number the other and then sprinkle that with dynamic nlet creation and you're done? Then you can simply abstract most cases.
On 2/14/2016 11:36 AM, Matt Barber wrote:
[gt~] is a great example of something that could work as an abstraction, except for the pesky right inlet which should take a signal if there's no creation argument, but float otherwise.
On Sun, Feb 14, 2016 at 10:50 AM, Ivica Bukvic ico@vt.edu wrote:
What I am also trying to do eventually in pd-l2ork is weed out redundant objects and only keep the ones that do the said task the best while still supporting other objects' idiosyncrasies (if any). There is absolutely no reason to have multiple objects of the same kind. Ultimately, one could keep all the externals in the same folder and completely do away with all the declares, imports, and other things that make learning pd unnecessarily harder. -- Ivica Ico Bukvic, D.M.A. Associate Professor Computer Music ICAT Senior Fellow Director -- DISIS, L2Ork Virginia Tech School of Performing Arts – 0141 Blacksburg, VA 24061 (540) 231-6139 ico@vt.edu www.performingarts.vt.edu disis.icat.vt.edu l2ork.icat.vt.edu ico.bukvic.net On Feb 14, 2016 8:40 AM, "Fred Jan Kraan" fjkraan@xs4all.nl wrote:
Hi Alexandre,
guess some of it is in: http://fjkraan.home.xs4all.nl/digaud/puredata/cyclone/cycloneToDo.html
This list is also becoming a list of what has been done.
As with _nettles_
"try to resurrect as independent object library"
Anyway, tell me if this gets includes on this file.
Yes, the nettles-objects are part of the latest cyclone versions. They are part of the nettles library, which can be loaded with [declare]. Not all operating systems like the '<' and '>' in the object names and there is overlap with other library objects, so only loading them when needed is cleaner.
cheers
ps. count me in for help with the help files
Great!
Greetings,
Fred Jan
2016-02-11 22:18 GMT-02:00 Alexandre Torres Porres <porres@gmail.com mailto:porres@gmail.com>:
Howdy, it's a known fact brazilians will start the year only after carnival, so here I am. I'd like to share my list of things to do with existing Cyclone Objetcs. Obviously there might be other issues with other objects that would make them up to date with the current version of Max (Max 7). Nonetheless, this is what I find relevant, and I've been really checking it through. It's only about 11 objects, some has already been discussed here and might have been fixed or in the process to be taken care of, forgive me if so. I have it attached and also as a link to a google dochttps://docs.google.com/document/d/1L_dUNgznfhaZHPKMJ3jJ_p9uIXRVP6Rs9-3nXy2Q...
Next, I will get together a list of new objects I think should be included, many of which I've already made as abstractions (kind of to show how it works like I did with [teeth~], cause I really think they should all be done as externals). Cheers
Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list http://lists.puredata.info/listinfo/pd-list
Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list http://lists.puredata.info/listinfo/pd-list
Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list