It would relatively easy to add a right outlet to [random]. Another option might be an explicit [seed] object which could give you further control or perhaps some creation flags for [random] as well.
you still have control on the seed... just seed it
no because you need to add an extra outlet to [random] and prints out the seed value.
or even seed it with the system time on creation?
if one adds now this behaviour one need to put a flag for backward compatibility.
what if [seed( without argument would take the current system time?
this could be an idea but one need the second outlet.
ciao -Marco Matteo Markidis
Dan Wilcox @danomatika http://twitter.com/danomatika danomatika.com http://danomatika.com/ robotcowboy.com http://robotcowboy.com/
wouldn´t it be more interesting/useful to incorporate a [date] object into vanilla pd, from which it would be trivial to generate unique seeds, but which also could be used in (many) other contexts?
hans
Am 31.05.2018 um 19:21 schrieb Dan Wilcox danomatika@gmail.com:
It would relatively easy to add a right outlet to [random]. Another option might be an explicit [seed] object which could give you further control or perhaps some creation flags for [random] as well.
you still have control on the seed... just seed it
no because you need to add an extra outlet to [random] and prints out the seed value.
or even seed it with the system time on creation?
if one adds now this behaviour one need to put a flag for backward compatibility.
what if [seed( without argument would take the current system time?
this could be an idea but one need the second outlet.
ciao -Marco Matteo Markidis
Dan Wilcox @danomatika danomatika.com robotcowboy.com
Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management -> https://lists.puredata.info/listinfo/pd-list
Yeah. I could use it as [date] is the only reason I have zexy installed right now.
On May 31, 2018, at 12:56 PM, hans w. koch hansw.koch@gmail.com wrote:
wouldn´t it be more interesting/useful to incorporate a [date] object into vanilla pd, from which it would be trivial to generate unique seeds, but which also could be used in (many) other contexts?
hans
Am 31.05.2018 um 19:21 schrieb Dan Wilcox danomatika@gmail.com:
It would relatively easy to add a right outlet to [random]. Another option might be an explicit [seed] object which could give you further control or perhaps some creation flags for [random] as well.
you still have control on the seed... just seed it
no because you need to add an extra outlet to [random] and prints out the seed value.
or even seed it with the system time on creation?
if one adds now this behaviour one need to put a flag for backward compatibility.
what if [seed( without argument would take the current system time?
this could be an idea but one need the second outlet.
ciao -Marco Matteo Markidis
Dan Wilcox @danomatika danomatika.com robotcowboy.com
Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management -> https://lists.puredata.info/listinfo/pd-list
Dan Wilcox @danomatika http://twitter.com/danomatika danomatika.com http://danomatika.com/ robotcowboy.com http://robotcowboy.com/
Warning: [date] won't work so well on Raspberry Pi startup scripts (no way to save date from boot to boot).
I think the best vanilla way on linux or mac is to read /dev/random into an array using soundfiler.
cheers Miller
On Thu, May 31, 2018 at 01:00:40PM -0500, Dan Wilcox wrote:
Yeah. I could use it as [date] is the only reason I have zexy installed right now.
On May 31, 2018, at 12:56 PM, hans w. koch hansw.koch@gmail.com wrote:
wouldnŽt it be more interesting/useful to incorporate a [date] object into vanilla pd, from which it would be trivial to generate unique seeds, but which also could be used in (many) other contexts?
hans
Am 31.05.2018 um 19:21 schrieb Dan Wilcox danomatika@gmail.com:
It would relatively easy to add a right outlet to [random]. Another option might be an explicit [seed] object which could give you further control or perhaps some creation flags for [random] as well.
you still have control on the seed... just seed it
no because you need to add an extra outlet to [random] and prints out the seed value.
or even seed it with the system time on creation?
if one adds now this behaviour one need to put a flag for backward compatibility.
what if [seed( without argument would take the current system time?
this could be an idea but one need the second outlet.
ciao -Marco Matteo Markidis
Dan Wilcox @danomatika danomatika.com robotcowboy.com
Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management -> https://lists.puredata.info/listinfo/pd-list
Dan Wilcox @danomatika http://twitter.com/danomatika danomatika.com http://danomatika.com/ robotcowboy.com http://robotcowboy.com/
Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management -> https://lists.puredata.info/listinfo/pd-list
I was thinking it would just return the posix date via outlets or a list. Why would Pd need to save the previous date?
On May 31, 2018, at 1:11 PM, Miller Puckette msp@ucsd.edu wrote:
Warning: [date] won't work so well on Raspberry Pi startup scripts (no way to save date from boot to boot).
I think the best vanilla way on linux or mac is to read /dev/random into an array using soundfiler.
cheers Miller
On Thu, May 31, 2018 at 01:00:40PM -0500, Dan Wilcox wrote:
Yeah. I could use it as [date] is the only reason I have zexy installed right now.
On May 31, 2018, at 12:56 PM, hans w. koch hansw.koch@gmail.com wrote:
wouldn´t it be more interesting/useful to incorporate a [date] object into vanilla pd, from which it would be trivial to generate unique seeds, but which also could be used in (many) other contexts?
hans
Am 31.05.2018 um 19:21 schrieb Dan Wilcox danomatika@gmail.com:
It would relatively easy to add a right outlet to [random]. Another option might be an explicit [seed] object which could give you further control or perhaps some creation flags for [random] as well.
you still have control on the seed... just seed it
no because you need to add an extra outlet to [random] and prints out the seed value.
or even seed it with the system time on creation?
if one adds now this behaviour one need to put a flag for backward compatibility.
what if [seed( without argument would take the current system time?
this could be an idea but one need the second outlet.
ciao -Marco Matteo Markidis
Dan Wilcox @danomatika danomatika.com robotcowboy.com
Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management -> https://lists.puredata.info/listinfo/pd-list
Dan Wilcox @danomatika http://twitter.com/danomatika danomatika.com http://danomatika.com/ robotcowboy.com http://robotcowboy.com/
Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management -> https://lists.puredata.info/listinfo/pd-list
Dan Wilcox @danomatika http://twitter.com/danomatika danomatika.com http://danomatika.com/ robotcowboy.com http://robotcowboy.com/
The Pi always boots at a constant date (no battery to keep a clock running).
cheers M On Thu, May 31, 2018 at 01:14:08PM -0500, Dan Wilcox wrote:
I was thinking it would just return the posix date via outlets or a list. Why would Pd need to save the previous date?
On May 31, 2018, at 1:11 PM, Miller Puckette msp@ucsd.edu wrote:
Warning: [date] won't work so well on Raspberry Pi startup scripts (no way to save date from boot to boot).
I think the best vanilla way on linux or mac is to read /dev/random into an array using soundfiler.
cheers Miller
On Thu, May 31, 2018 at 01:00:40PM -0500, Dan Wilcox wrote:
Yeah. I could use it as [date] is the only reason I have zexy installed right now.
On May 31, 2018, at 12:56 PM, hans w. koch hansw.koch@gmail.com wrote:
wouldnŽt it be more interesting/useful to incorporate a [date] object into vanilla pd, from which it would be trivial to generate unique seeds, but which also could be used in (many) other contexts?
hans
Am 31.05.2018 um 19:21 schrieb Dan Wilcox danomatika@gmail.com:
It would relatively easy to add a right outlet to [random]. Another option might be an explicit [seed] object which could give you further control or perhaps some creation flags for [random] as well.
> you still have control on the seed... just seed it
no because you need to add an extra outlet to [random] and prints out the seed value.
> or even seed it with the system time on creation?
if one adds now this behaviour one need to put a flag for backward compatibility.
> what if [seed( without argument would take the current system time?
this could be an idea but one need the second outlet.
ciao -Marco Matteo Markidis
Dan Wilcox @danomatika danomatika.com robotcowboy.com
Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management -> https://lists.puredata.info/listinfo/pd-list
Dan Wilcox @danomatika http://twitter.com/danomatika danomatika.com http://danomatika.com/ robotcowboy.com http://robotcowboy.com/
Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management -> https://lists.puredata.info/listinfo/pd-list
Dan Wilcox @danomatika http://twitter.com/danomatika danomatika.com http://danomatika.com/ robotcowboy.com http://robotcowboy.com/
Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management -> https://lists.puredata.info/listinfo/pd-list
Ok, makes sense. A [date] object would still be useful for my case: generating filenames with timestamps. :)
That's why I was thinking of some sort of [seed] or [salt] object which would wrap reading from a default pseudo-random source such as /dev/random or some system equivalent.
On May 31, 2018, at 1:16 PM, Miller Puckette msp@ucsd.edu wrote:
The Pi always boots at a constant date (no battery to keep a clock running).
cheers M On Thu, May 31, 2018 at 01:14:08PM -0500, Dan Wilcox wrote:
I was thinking it would just return the posix date via outlets or a list. Why would Pd need to save the previous date?
On May 31, 2018, at 1:11 PM, Miller Puckette msp@ucsd.edu wrote:
Warning: [date] won't work so well on Raspberry Pi startup scripts (no way to save date from boot to boot).
I think the best vanilla way on linux or mac is to read /dev/random into an array using soundfiler.
cheers Miller
On Thu, May 31, 2018 at 01:00:40PM -0500, Dan Wilcox wrote:
Yeah. I could use it as [date] is the only reason I have zexy installed right now.
On May 31, 2018, at 12:56 PM, hans w. koch hansw.koch@gmail.com wrote:
wouldn´t it be more interesting/useful to incorporate a [date] object into vanilla pd, from which it would be trivial to generate unique seeds, but which also could be used in (many) other contexts?
hans
Am 31.05.2018 um 19:21 schrieb Dan Wilcox danomatika@gmail.com:
It would relatively easy to add a right outlet to [random]. Another option might be an explicit [seed] object which could give you further control or perhaps some creation flags for [random] as well.
>> you still have control on the seed... just seed it > > no because you need to add an extra outlet to [random] and prints out the > seed value. > >> or even seed it with the system time on creation? > > if one adds now this behaviour one need to put a flag for backward > compatibility. > >> what if [seed( without argument would take the current system time? > > this could be an idea but one need the second outlet. > > ciao > -Marco Matteo Markidis
Dan Wilcox @danomatika danomatika.com robotcowboy.com
Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management -> https://lists.puredata.info/listinfo/pd-list
Dan Wilcox @danomatika http://twitter.com/danomatika danomatika.com http://danomatika.com/ robotcowboy.com http://robotcowboy.com/
Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management -> https://lists.puredata.info/listinfo/pd-list
Dan Wilcox @danomatika <http://twitter.com/danomatika http://twitter.com/danomatika> danomatika.com http://danomatika.com/ <http://danomatika.com/ http://danomatika.com/> robotcowboy.com http://robotcowboy.com/ <http://robotcowboy.com/ http://robotcowboy.com/>
Pd-list@lists.iem.at mailto:Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management -> https://lists.puredata.info/listinfo/pd-list https://lists.puredata.info/listinfo/pd-list
Dan Wilcox @danomatika http://twitter.com/danomatika danomatika.com http://danomatika.com/ robotcowboy.com http://robotcowboy.com/
but couldn´t that pi limitation worked around by a loadbang -delay combo to read a date, once the system has established one? would need mention in the helpfile though.
currently i use [shell] to read a date/hour into pd. it works well (e.g. in an installation, where i cue this every second, running since mid march), but always wondered, why there was no native way. time stamped filenames would be another good use, among many others. i even made a donation once, to have this in mobmuplat.
cheers hans
Am 31.05.2018 um 20:20 schrieb Dan Wilcox danomatika@gmail.com:
Ok, makes sense. A [date] object would still be useful for my case: generating filenames with timestamps. :)
That's why I was thinking of some sort of [seed] or [salt] object which would wrap reading from a default pseudo-random source such as /dev/random or some system equivalent.
On May 31, 2018, at 1:16 PM, Miller Puckette msp@ucsd.edu wrote:
The Pi always boots at a constant date (no battery to keep a clock running).
cheers M On Thu, May 31, 2018 at 01:14:08PM -0500, Dan Wilcox wrote:
I was thinking it would just return the posix date via outlets or a list. Why would Pd need to save the previous date?
On May 31, 2018, at 1:11 PM, Miller Puckette msp@ucsd.edu wrote:
Warning: [date] won't work so well on Raspberry Pi startup scripts (no way to save date from boot to boot).
I think the best vanilla way on linux or mac is to read /dev/random into an array using soundfiler.
cheers Miller
On Thu, May 31, 2018 at 01:00:40PM -0500, Dan Wilcox wrote:
Yeah. I could use it as [date] is the only reason I have zexy installed right now.
On May 31, 2018, at 12:56 PM, hans w. koch hansw.koch@gmail.com wrote:
wouldn´t it be more interesting/useful to incorporate a [date] object into vanilla pd, from which it would be trivial to generate unique seeds, but which also could be used in (many) other contexts?
hans
> Am 31.05.2018 um 19:21 schrieb Dan Wilcox danomatika@gmail.com: > > It would relatively easy to add a right outlet to [random]. Another option might be an explicit [seed] object which could give you further control or perhaps some creation flags for [random] as well. > >>> you still have control on the seed... just seed it >> >> no because you need to add an extra outlet to [random] and prints out the >> seed value. >> >>> or even seed it with the system time on creation? >> >> if one adds now this behaviour one need to put a flag for backward >> compatibility. >> >>> what if [seed( without argument would take the current system time? >> >> this could be an idea but one need the second outlet. >> >> ciao >> -Marco Matteo Markidis > > -------- > Dan Wilcox > @danomatika > danomatika.com > robotcowboy.com > > > > _______________________________________________ > Pd-list@lists.iem.at mailing list > UNSUBSCRIBE and account-management -> https://lists.puredata.info/listinfo/pd-list
Dan Wilcox @danomatika http://twitter.com/danomatika danomatika.com http://danomatika.com/ robotcowboy.com http://robotcowboy.com/
Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management -> https://lists.puredata.info/listinfo/pd-list
Dan Wilcox @danomatika http://twitter.com/danomatika danomatika.com http://danomatika.com/ robotcowboy.com http://robotcowboy.com/
Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management -> https://lists.puredata.info/listinfo/pd-list
Dan Wilcox @danomatika danomatika.com robotcowboy.com
On Thu, May 31, 2018 at 3:09 PM, hans w. koch hansw.koch@gmail.com wrote:
but couldn´t that pi limitation worked around by a loadbang -delay combo to read a date, once the system has established one? would need mention in the helpfile though.
The pi might not be connected to any network, in which case it will always
start at Jan 1, 2000. /dev/random may not be usable right at startup as it needs time to accumulate entropy, .dev/urandom is guaranteed to give some sort of random. Reading a few bytes from /dev/urandom into a table then combining them into a float or long int seems like a better idea. There's an equivalent method in the WIndows API (SystemPrng), but it would have to be coded into an external since it's not a file like in linux.
Martin
Hello,
I find it useful to wrap random in an abstraction, so that I use $2 as an instance id, to both receive a "seed" and to add the abstraction id to the seed.
But, it would be really useful to have a unique instance id already generated in the [random] object, and a global symbol to bind all [random]s to send a seed and increment it by each unique id.
what do you think? is this possible?
I tested briefly and thie wrapping method really gives different random streams provided there is a different seed when opening the patch
cheers,
fede
On May 31, 2018, at 10:00 PM, Martin Peach chakekatzil@gmail.com wrote:
On Thu, May 31, 2018 at 3:09 PM, hans w. koch hansw.koch@gmail.com wrote: but couldn´t that pi limitation worked around by a loadbang -delay combo to read a date, once the system has established one? would need mention in the helpfile though.
The pi might not be connected to any network, in which case it will always start at Jan 1, 2000. /dev/random may not be usable right at startup as it needs time to accumulate entropy, .dev/urandom is guaranteed to give some sort of random. Reading a few bytes from /dev/urandom into a table then combining them into a float or long int seems like a better idea. There's an equivalent method in the WIndows API (SystemPrng), but it would have to be coded into an external since it's not a file like in linux.
Martin
Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management -> https://lists.puredata.info/listinfo/pd-list
On 06/30/2018 07:51 PM, Fede Camara Halac wrote:
Hello,
I find it useful to wrap random in an abstraction, so that I use $2 as an instance id, to both receive a "seed" and to add the abstraction id to the seed.
But, it would be really useful to have a unique instance id already generated in the [random] object, and a global symbol to bind all [random]s to send a seed and increment it by each unique id.
what do you think? is this possible?
i think it is possible, and i don't think it should be implemented. a) it's super-easy to implement as an abstraction. what would a built-in solution give you that an abstraction cannot give you? b) it adds a false sense of randomness. either the seed should be stupid enough (as is now), or it should have proper entropy. your solution is inbetween.
gfmsadf IOhannes
a) it's super-easy to implement as an abstraction. what would a built-in solution give you that an abstraction cannot give you?
well, I think it would guarantee that there is uniqueness on each random object, independently on pd instances ideally
b) it adds a false sense of randomness. either the seed should be stupid enough (as is now), or it should have proper entropy. your solution is inbetween.
I agree, yes. Proper entropy or something close to that would mean a proper solution. I just want to be practical and efficient with what we have
Cheers!
f
- Fede Camara Halac camarafede@gmail.com [2018-06-30 19:51]:
Hello,
I find it useful to wrap random in an abstraction, so that I use $2 as an instance id, to both receive a "seed" and to add the abstraction id to the seed.
Perhaps using $0 might be better then as it is unique and automatically assigned.
Perhaps using $0 might be better then as it is unique and automatically assigned.
But $0 itself is deterministic--the first instance is 1003, the second is 1004, etc.
- Liam Goodacre liamg_uw@hotmail.com [2018-07-02 06:34]:
Perhaps using $0 might be better then as it is unique and automatically assigned.
But $0 itself is deterministic--the first instance is 1003, the second is 1004, etc.
Yes, but it will provide a different seed to [random] inside every abstraction. Forgive me if I understand your intentions in the wrong way-
Each new instance of [random] already seeds itself differently, so adding $0 isn't going to change anything there. The problem is that the seeds are repeated when you restart PD, so you'll get the same pseudo-random results every time you open the program. $0 won't help here, because the $0 values also repeat when you restart PD. ________________________________ From: Pd-list pd-list-bounces@lists.iem.at on behalf of Peter P. peterparker@fastmail.com Sent: 02 July 2018 18:26 To: pd-list@lists.iem.at Subject: Re: [PD] Random
- Liam Goodacre liamg_uw@hotmail.com [2018-07-02 06:34]:
Perhaps using $0 might be better then as it is unique and automatically assigned.
But $0 itself is deterministic--the first instance is 1003, the second is 1004, etc.
Yes, but it will provide a different seed to [random] inside every abstraction. Forgive me if I understand your intentions in the wrong way-
Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management -> https://lists.puredata.info/listinfo/pd-list
-
Dan Wilcox -
Fede Camara Halac -
hans w. koch -
IOhannes m zmölnig -
Liam Goodacre -
Martin Peach -
Miller Puckette -
Peter P.