Fritz Pape

I want to be able to control various float boxes with a MIDI device w/ rotary knobs (like the Behringer BCR2000 or that Doepfer Pocket Dial or...anything better?). I have a few patches that do various droney things in small increments and would just love to be able to slowly and accurately raise/lower those numbers by as small of an increment as I can (like, 0.00001 in either direction per turn). Is this possible? Or am I stuck to the 0-127 thing?

This is a whole series of questions;
I want to be able to control various float boxes with a MIDI device w/ rotary knobs (like the Behringer BCR2000 or that Doepfer Pocket Dial or...anything better?). I have a few patches that do various droney things in small increments and would just love to be able to slowly and accurately raise/lower those numbers by as small of an increment as I can (like, 0.00001 in either direction per turn). Is this possible? Or am I stuck to the 0-127 thing?


rotary-knobs-for-tiny-numbers

Hi Fritz,
The main thing to check is if the rotary knobs do infinite rotation, and what it outputs after it hits 0 or 127. I mean, if you keep turning it does it stil outputs 127 or does it stop, or starts from 0 gain? or something else?
If the knobs turn infinitly you can!
You could start from somthing like this (this assumes that the knob repeatedly outputs 127 and 0 at its "ends" when you keep turning it):

You can configure the midi controller to send midi cc in relative mode, e.g. value >  64 when turning clockwise and increment the value yourself.

Or your midi controller may be able to send out 14-bit resolution midi cc messages (like faderfox's controllers), This will give you values ranging from 0 to 16383.


given that the condition nikolas k is talking about is met (controller sends the 0 and the 127 again if you hit the limit) the following expression will do what you need:

expr ( ( ((($f1!=0) && ($f1!=127) && ($f1>$f2)) || ($f1==127)) * 1. ) + ( ((($f1!=0) && ($f1!=127) && ($f1

where f2 is the last f1 and f3 a variable factor for the step size.

p.s. note that the doepfer endless dial is discontinued long ago, while the behringer comes with modified firmware and various user presets which (afaik) allows that mode you need among many others.


Rotary knobs for tiny numbers?

demers

Can anyone knowledgeable in Javascript show me how to access some nested dictionaries in a JSON file?

Initially the file had an unnamed array of subdictionaries with identical groups of keys (see attached), but dict wouldn't accept it so I enclosed it in a main key.

The JS code I have (below) gets me the name, key and subdictionary names (named 0, 1, 2... since they are actually unnamed). But when I try to go one step further into the subdictionary key-value pairs, I just get 'jsobject 0' in the Max console. Object.keys doesn't seem to work.

Also, what about querying values tied to a single key that recurs in all subdictionaries?

    dict.import_json("fruits and peaches json3.json");

Can anyone knowledgeable in Javascript show me how to access some nested dictionaries in a JSON file? 

function get_keys()
{
	var dict = new Dict;
	dict.import_json("fruits and peaches json3.json");
	var keys = dict.getkeys();
	var name = dict.name;
	post("dict name: ");
	post(name);
	post();
	post("key: ");
	post(keys);
	post();
        var arr = dict.get(keys);
	post("subkeys: ");
	var arr2 = Object.keys(arr);
	post(arr2);
         var arr3 = dict.get(keys[1]);
	post();
	post(arr3);
}


javascript-to-access-nested-keys-by-category

Firstly, when getting the keys from a dictionary in JS, you should check if the keys you got are an array or a string.
If the dict has only one key, the dict.getkeys() will return a string, if it has more it will return an array of strings.
What I do to prevent this is the following:

var dict = new Dict("my_dictionary_mame");
var keys = dict.getKeys();

// This will force the "keys" to be an array, even with a single element.
if (typeof(keys ) == "string")
		keys = [keys ];

Other than that, when getting the subdict your line is var arr = dict.get(keys);.
If the keys are an array, it will return you nothing.
You should do it like var arr = dict.get(keys[0]); or any other index you want.
Now, to get the keys of a "subdict", you do it the same way:

var dict = new Dict("my_dict_bla_bla");
var keys = dict.getkeys();
if (typeof(keys ) == "string")
		keys = [keys ];

// Here you get the subdictionary keys
var subdict = dict.get(keys[0]); // This is the first element of the topdict. Lets say it is a dictionary itself. 
var subkeys = subdict.getkeys();

//etc...

Really appreciate the help. I'm still running into trouble though; typeof identifies an object, but not whether it is an array or not. Oddly enough, even with the method you mentioned (converting a value subkey to [subkey]), Max says it isn't an array, but then generates a message saying it's not an object either! 

Also, a command like: post(dict.fruitspeaches[0].business);
...doesn't yield anything, though I've tried this in a Javascript console, and it works.

The basic structure of the JSON file is below. The code is right after. Appreciate any clarification!

{
	"fruitspeaches" : [
		{
			"business" : "Francis Motuzick Jr",
			"category" : "Fruit",
			"farmer_id" : "3402",
			"item" : "Peaches",
			"l" : "0",
			"location_1" : 			{
				"type" : "Point",
				"coordinates" : [ -73.096275, 41.779892 ]
			}
		}
...

	var dict = new Dict;
	dict.import_json("fruits and peaches 3.json");
	var keys = dict.getkeys();
	if(typeof(keys) == "string") // Force the 'keys' to be an array, even with single element
		keys = [keys];
	var name = dict.name;
	post("dict name: ", name);
	post("key: ", keys);
	post("dict type: ", typeof(dict), "array? ", Array.isArray(dict)); // object, not an array
	post("key type: ", typeof(keys), "array? ", Array.isArray(keys)); // object, is an array
	post("keys[0] type: ", typeof(keys[0]), "array? ", Array.isArray(keys[0])); // says it's a string
	post();
	var subdict = new Dict(keys[0]);
	if(typeof(subdict) == "string")
		subdict = [subdict];
	post("subdict type: ", typeof(subdict), "array? ", Array.isArray(subdict));
	post();
	var subkeys = subdict.getkeys();
	post("subkeys type: ", typeof(subkeys), "array? ", Array.isArray(subkeys));
	post();
	post(Object.getOwnPropertyNames(subkeys));
	post();
	if(typeof(subkeys) == "string")
		subkeys = [subkeys];
	post(subkeys);
	post();


Ok, I think there are some misconceptions that lead you to some errors in handlig the dictionaries in SJ.
You main proble is in this code:

var subdict = new Dict(keys[0]);
if(typeof(subdict) == "string")
subdict = [subdict];

.
the var subdict = new Dict(keys[0]) will create a new dict that will be named as the first key of the first Dict. That doesn't mean that it is the a sub Dict. The correct way to get the sub dict is by the get("keyName") method of dict like so:
var subdict = dict.get(keys[0]);. This works because thekeys[0]` will return (in this case) the name of the 1st key, that happens to be another dictionary. To make sure that the dictionary element you get is another dict you should check that is it object but not array.
Secondly, here: if(typeof(subdict) == "string") subdict = [subdict]; you are comparing the dict element to see if it is a string, not its keys.
So it will return an array with on object inside, that is in this the sub dictionary. 

So, to get the keys you do var mykeys = mydict.getkeys(). This will return either an array of string, or a string (not ver cool but it is the way it is...). Only here you should do this check if(typeof(mykeys ) == "string") subdict = [mykeys ];, and that is so you can easier write that handles the keys i.e. be sure that it is an array, and not have to check every time if its not.
To get a subdictionary you should do var thesubdict = mydict.get(keys[N]);. Here you should make sure that it is indeed a dictionary - unless you are 100% sure that the Nth key is a subdictionary, bu still...
the var thesubdict (or whatever name) you should treat as a dict (because it is a dict!), meaning the way you got the first dicts keys etc, the same way you will for the subdict, like var sudictkeys = thesubdict .getkeys() etc...

I made the above corrections to your code and it works:

var dict = new Dict;
dict.import_json("fruits and peaches 3.json");

var keys = dict.getkeys();
if(typeof(keys) == "string") // Force the ‘keys’ to be an array, even with single element
keys = [keys];

var name = dict.name;
post("dict name: ", name);
post("\nkey: ", keys);
post("\ndict type: ", typeof(dict), "array? ", Array.isArray(dict)); // object, not an array
post("\nkey type: ", typeof(keys), "array? ", Array.isArray(keys)); // object, is an array
post("\nkeys[0] type: ", typeof(keys[0]), "array? ", Array.isArray(keys[0])); // says it’s a string
post();
post();

var subdict = dict.get(keys[0]);
post(subdict.stringify());
post("subdict type: ", typeof(subdict), "array? ", Array.isArray(subdict));
post();

var subkeys = subdict.getkeys();
post("subkeys type: ", typeof(subkeys), "array? ", Array.isArray(subkeys));
post(subkeys);
post();
post(Object.getOwnPropertyNames(subkeys));
post();
post();

if(typeof(subkeys) == "string")
subkeys = [subkeys];
post(subkeys);
post();


also the json you posted wasn't formatted correctly, so here it is
(also make sure that the json you use is correctly formated. To test it you can load it in an object [dict], if it is OK it will be read, else it will not) :

Very much appreciate the explanation on how to invoke the subdict and proper formatting!
Strangely though, 2 problems came up with the corrected code you posted:

1. I get an error message that subdict.stringify is not a function.
2. The JSON formatting you did (with curly brackets throughout) wasn't accepted by my dict object, only the one with alternating { [ { was.

To be clear, the snippet of JSON I posted was only the beginning, a better representation would be this:

Javascript to access nested keys by category

Annie Jiang

Just wondering if you knew how to adjust the volume via tilting?

change-volume-via-tilting

Generaly peaking, you can probably do almost everything, but unless you give a bit more info, can't be sure.
I assume you probably mean to tilt a phone/table (?) and control some parameter (this case the volume) in a max pathcer, then yes!!
-N


Change volume via tilting

thetechnobear

so a while back I developed an external for Max6, and went thru the pain of working out the build system.

Im back again... and now trying with the Max7 SDK... and realised Ive forgotten most of it :)

(this is on macOS 10.12, with latest Xcode, and latest max7 sdk from GitHub)

Ive no issue with the code just making it build :)

the sdk help files is great for explaining the API, but really doesnt talk at all about setting up the build environment :(

so Ive a couple of questions, id really appreciate it if others could help me with....

- is there a C++ interface ? or do I still have to have a C wrapper, to then hit my C++ objects

copy maxmspdsk.xcconfig , and then edit it to reflect your environment (e.g. sdk location)

ive tried to alter the link link in the xcconfig file, but it seems to be ignored eg

OTHER_LDFLAGS = -L$(SRCROOT)/../build/release/lib -framework MaxAudioAPI -framework JitterAPI $(C74_SYM_LINKER_FLAGS) -lmec-api -Wl,-rpath,$(SRCROOT)/../build/release/lib

the whole maxmspsdk.xcconfig, seems to be very complex compared to just using makefiles...

I'll admit I dont use Xcode much, more makefiles/cmake .. has anyone used cmake on a max external project that would be so much easier to manage .

or have I missed some document, which explains how to create a max external from scratch (simple) that you dont put in the SDK source folder!

so a while back I developed an external for Max6, and went thru the pain of working out the build system.
Im back again... and now trying with the Max7 SDK... and realised Ive forgotten most of it :)

(this is on macOS 10.12, with latest Xcode, and latest max7 sdk from GitHub) 

Ive no issue with the code just making it build :)
the sdk help files  is great for explaining the API, but really doesnt talk at all about setting up the build environment :( 

- is there a C++ interface ? or do I still have to have a C wrapper, to then hit my C++ objects 

- for xcode, is the correct process...
    copy maxmspdsk.xcconfig , and then edit it to reflect your environment (e.g. sdk location)

- how do you link in your own externals?
ive tried to alter the link link in the xcconfig file, but it seems to be ignored eg 

the whole maxmspsdk.xcconfig, seems to be very complex compared to just using makefiles...
I'll admit I dont use Xcode much, more makefiles/cmake .. has anyone used cmake on a max external project that would be so much easier to manage .

or have I missed some document, which explains how to create a max external from scratch (simple) that you dont put in the SDK source folder!


developing-external-for-max7

ok, Ive just stumbled on the max-api and max-devkit

are these replacing the max-sdk, since the readme refers to the sdk as the 'legacy max sdk'

this api looks much more interesting to me, given the build is via cmake, and appears to be 'cross platform' rather than maintaining a project file for each platform - also it looks quite low-level, which might suit my project (which is integrating 3rd party hardware controllers) 

Im building something new, and doesn't need much to integrate with max, this looks like a promising direction.
however, I can find little documentation on it, or comments on using max-api vs max-sdk.

the readme mentions the sdk has more 'features', does this mean it is in its early stages to eventually replace the sdk? but perhaps is not complete yet?

is there a document discussing the pros/cons? or when to use it vs the max-sdk?


posting in solidarity with this thread - the documentation for the code base itself is fine but the readme for the Max-DevKit is abysmal. Eric Lyon's text on externals makes building for PD a breeze (


the max-devkit uses git submodules to wrangle in max-api -- make sure you source both projects when attempting the walkthrough. if you prefer to download zips from github, make sure you download both max-devkit and max-api zip files and unzip them. you'll have to rename max-apiWHATEVER to plain max-api, then drag it into the max-devkit/source overriding the empty folder by the same name.

if you prefer to use git clone, clone max-devkit, then complete the following to source max-api submodule:

either approach gives you all the bits you need -- now just follow your platform's build steps per the walkthrough and your tasty new reference externals will be sitting in the max-devkit/externals folder ready to slide into your Max path

developing external for max7

mahoku

i am trying to figure out how to control the start and stop of a recording of an IP camera (HIKvision) for a research project. let me explain: I am using Max to control 3 TVs where i display images to a dolphin. the dolphin has then to choose certain images and press a response paddle. the whole setup is controlled via Max where a preloaded schedule of images is run through and the responses of the dolphin is recorded via max (inc reaction time etc). so far so good. i figured out how make the whole thing work and i can 24 trials and everything seems to be fine.

but- and here is my problem - i have a new IP camera that installed across the pool of the dolphin and records the research trials. what i would like to is to automatically start recording when a new trial is selected then record until the dolphin has made her choice - and then automatically stop. the camera allows me to record to a server and it does have RTSP - but i have looked and so far failed to figure out how to send the control to the camera. right now i have to go into Safari and load the web plugin of the the camera and then switch between max and Safari all the time - this of course leads to missed trials or erroneous recordings...

So how do get start the camera recording automatically when i run a trial? and then stop it once a paddle is pressed?

thank to anyone who can point me in the right direction....

Hi-
i am trying to figure out how to control the start and stop of a recording of an IP camera (HIKvision) for a research project. let me explain: I am using Max to control 3 TVs where i display images to a dolphin. the dolphin has then to choose certain images and press a response paddle. the whole setup is controlled via Max where a preloaded schedule of images is run through and the responses of the dolphin is recorded via max (inc reaction time etc). so far so good. i figured out how make the whole thing work and i can 24 trials and everything seems to be fine.
but- and here is my problem - i have a new IP camera that installed across the pool of the dolphin and records the research trials. what i would like to is to automatically start recording when a new trial is selected then record until the dolphin has made her choice - and then automatically stop. the camera allows me to record to a server and it does have RTSP - but i have looked and so far failed to figure out how to send the control to the camera. right now i have to go into Safari and load the web plugin of the the camera and then switch between max and Safari all the time - this of course leads to missed trials or erroneous recordings...
So how do get start the camera recording automatically when i run a trial? and then stop it once a paddle is pressed?
i am at a loss....
thank to anyone who can point me in the right direction....

control-of-an-ip-camera-via-max

Have you tried loading the camera's interface using Max's jweb object?


no-the camera interface is a web-plugin that only works with Safari - and any documentation that i have is in Chinese...
so what i would like to do is teh following:
1) get a camera view to display on my max patcher. (i have not even gotten that far)
2) start and stop the camera recording (at 1080) to a server - right now i need to do that manually for each trial through safari.... but i always loose track and since the trials are very quick its easy to mess up... 

i will try exploring jweb (never used that before)

hm - jweb does not work either - it does correctly load the webpage - but not the plugin - so i can't see the camera.


Control of an IP camera via Max

Thijs Koerselman

I developed an external which connects to a streaming TCP/IP socket. In

order to receive the (time-critical) data as fast as possible, the object

has to handle the receiving and parsing of the data in a separate thread,

otherwise the max scheduler locks up. I have instantiated a Posix thread to

receive and parse the data, which writes to a shared data buffer in my

object. At the same time a Max clock is running (in the main scheduler) for

reading the buffer and writing the data to an outlet.

I use a circular buffer to pass the data between the Posix thread and the

clock function. I first used std::vector for it, but it didn't like that at

all (debug assertion failure related to iterators), so now I use a plain

array which seems to be ok. The object works fine for the time being, but

I'm pretty sure what I'm doing it not thread safe.

Can I make use of critical regions or mutexes to protect the data buffer? Is

there a Max native way to spawn a new thread and use critical regions with

it? I've read the section about critical regions, but I'm just not sure if

"So unless you are completely sure of what you are doing, and absolutely

need to make use of multiple critical regions to protect your code, we

suggest you use the global critical region."

So yeah, I'm clearly not completely sure:)

In earlier attempts I tried using the scheduler_set() and family to create a

new scheduler instead of using Posix, but this got me into trouble more than

anything. I find the documentation about this is subject very sparse at

best, and a search in the archives didn't turn up anything either. The text

in the paragraph "creating schedulers" in the writingexternals.pdf seems to

I developed an external which connects to a streaming TCP/IP socket. In
order to receive the (time-critical) data as fast as possible, the object
has to handle the receiving and parsing of the data in a separate thread,
otherwise the max scheduler locks up. I have instantiated a Posix thread to
receive and parse the data, which writes to a shared data buffer in my
object. At the same time a Max clock is running (in the main scheduler) for
reading the buffer and writing the data to an outlet.

I use a circular buffer to pass the data between the Posix thread and the
clock function. I first used std::vector for it, but it didn't like that at
all (debug assertion failure related to iterators), so now I use a plain
array which seems to be ok. The object works fine for the time being, but
I'm pretty sure what I'm doing it not thread safe.

Can I make use of critical regions or mutexes to protect the data buffer? Is
there a Max native way to spawn a new thread and use critical regions with
it? I've read the section about critical regions, but I'm just not sure if
it applies to my situation.

Quote:
"So unless you are completely sure of what you are doing, and absolutely
need to make use of multiple critical regions to protect your code, we
suggest you use the global critical region."

In earlier attempts I tried using the scheduler_set() and family to create a
new scheduler instead of using Posix, but this got me into trouble more than
anything. I find the documentation about this is subject very sparse at
best, and a search in the archives didn't turn up anything either. The text
in the paragraph "creating schedulers" in the writingexternals.pdf seems to
be messed up.

multi-threading-in-externals

On 2008 May 28, at 9:43 AM, Thijs Koerselman wrote:

> Dear List,
>
> I developed an external which connects to a streaming TCP/IP socket.  
> In order to receive the (time-critical) data as fast as possible,  
> the object has to handle the receiving and parsing of the data in a  
> separate thread, otherwise the max scheduler locks up. I have  
> instantiated a Posix thread to receive and parse the data, which  
> writes to a shared data buffer in my object. At the same time a Max  
> clock is running (in the main scheduler) for reading the buffer and  
> writing the data to an outlet.

> Can I make use of critical regions or mutexes to protect the data  
> buffer? Is there a Max native way to spawn a new thread and use  
> critical regions with it? I've read the section about critical  
> regions, but I'm just not sure if it applies to my situation.

It seems like you actually have two issues here: one is working with  
another thread, and the other is about thread-safety.

There is a Max native way to create and use new threads.  You can find  
the relevant coding support in ext_systhread.h.  I don't think it's  
really documented yet.

In Tap.Tools 3 there is a new object (tap.svn) which is a Subversion  
client wrapped up as a Max external.  This is still a work in  
progress, and the code is a bit of a mess, but it does use the  
systhread API defined in Max.  I've attached the source in the hope it  
will be useful.  Here are the relevant bits:

1. In the struct:
	t_systhread		thread;
	short			thread_running;

2. In the free routine, you need to make sure that you don't hang Max:
	if(x->thread_running > 0)
		systhread_join(x->thread, NULL);// avoid deadlock if thread should  
somehow be dead.

3. Down in maxsvn_anything() we our flag in the struct and call a  
method, which executes in a new thread (whose handle we are storing in  
our instance):
	x->thread_running = 1;
	systhread_create((method)maxsvn_doanything, x, 0, 0, 0, &x->thread);

4. When maxsvn_doanything() is done, and thus we don't need the thread  
anymore, we do this:
	x->thread_running = 0;
	systhread_exit(0);

In your case, you might create the thread with method that implements  
an infinite loop (I'm not entire sure what you are doing).  If you do,  
you may also want to make sure that your thread is throttled somehow,  
such as using systhread_sleep().  Something like this:
	while(1){
		// look to see if there is something to do
		// if so, then do it...

		// relinquish the processor briefly so Max can do the other things  
it needs to do
		systhread_sleep(25);
	}

> It seems like you actually have two issues here: one is working with
> another thread, and the other is about thread-safety.
>
> There is a Max native way to create and use new threads.  You can find the
> relevant coding support in ext_systhread.h.  I don't think it's really
> documented yet.
>
> In Tap.Tools 3 there is a new object (tap.svn) which is a Subversion client
> wrapped up as a Max external.  This is still a work in progress, and the
> code is a bit of a mess, but it does use the systhread API defined in Max.
>  I've attached the source in the hope it will be useful.  Here are the
> relevant bits:
>
>
Hi Tim,

Thanks a lot! That's exactly what I need. It looks really similar to what I
do with Posix atm. I'm glad I can get rid of that, because it was adding an
extra dependency since its not natively supported on windows.

I do use an infinite loop, but I don't need it to sleep. It's automatically
waiting for the socket to receive data, and I need to parse as soon as I get
it. It works fine like this and hardly uses any cpu so I'm happy.

You didn't mention critical regions, so just to be clear; I can create a
mutex by using the critical enter/exit with thread 0, right? Or is this
something I have to create my own critical region struct for?

On 2008 May 28, at 4:32 PM, Thijs Koerselman wrote:

>
> 2008/5/28 Timothy Place :
> It seems like you actually have two issues here: one is working with  
> another thread, and the other is about thread-safety.
>
> There is a Max native way to create and use new threads.  You can  
> find the relevant coding support in ext_systhread.h.  I don't think  
> it's really documented yet.
>
> In Tap.Tools 3 there is a new object (tap.svn) which is a Subversion  
> client wrapped up as a Max external.  This is still a work in  
> progress, and the code is a bit of a mess, but it does use the  
> systhread API defined in Max.  I've attached the source in the hope  
> it will be useful.  Here are the relevant bits:
>
>
> Hi Tim,
>
> Thanks a lot! That's exactly what I need. It looks really similar to  
> what I do with Posix atm. I'm glad I can get rid of that, because it  
> was adding an extra dependency since its not natively supported on  
> windows.
>
> I do use an infinite loop, but I don't need it to sleep. It's  
> automatically waiting for the socket to receive data, and I need to  
> parse as soon as I get it. It works fine like this and hardly uses  
> any cpu so I'm happy.
>
> You didn't mention critical regions, so just to be clear; I can  
> create a mutex by using the critical enter/exit with thread 0,  
> right? Or is this something I have to create my own critical region  
> struct for?

Great - I'm glad that was helpful.  Yes, you can use the global  
critical region (e.g critical_enter(0) and critical_exit(0)).  You  
could do something more to create your own private mutex if wanted,  
but you don't need to.

as mentioned before in this thread I took advantage of the possibility to create a thread inside an external.
Thanks for all the information! It works and was really easy to implement.
But I cannot share the experience, that it is not using so much cpu: in my case the cpu in
the activity-monitor of my Mac is going from 3.1 up to 9. The 'Dsp Status' keeps its 0 (no audio here).
Is there any improvement in my code?
(Mac Pro 2.5 GHz Core 2 Duo 4 GB Ram 10.6.4)
Thanks in advance for any help!

#include "ext.h"
#include "ext_obex.h"
#include "ext_strings.h"
#include "ext_common.h"
#include "ext_systhread.h"

using namespace std;

// a wrapper for cpost() only called for debug builds on Windows
// to see these console posts, run the DbgView program (part of the SysInternals package distributed by Microsoft)
#if defined( NDEBUG ) || defined( MAC_VERSION )
#define DPOST
#else
#define DPOST cpost
#endif

// a macro to mark exported symbols in the code without requiring an external file to define them
#ifdef WIN_VERSION
	// note that this is the required syntax on windows regardless of whether the compiler is msvc or gcc
	#define T_EXPORT __declspec(dllexport)
#else // MAC_VERSION
	// the mac uses the standard gcc syntax, you should also set the -fvisibility=hidden flag to hide the non-marked symbols
	#define T_EXPORT __attribute__((visibility("default")))
#endif

// max object instance data
typedef struct _myclocker {
	t_object			c_box;
	void				*c_outlet;
	t_systhread	 thread;
	short	 thread_running;
} t_myclocker;

// prototypes
void*	myclocker_new(t_symbol *s, long argc, t_atom *argv);
void	myclocker_free(t_myclocker* x);
void	myclocker_assist(t_myclocker *x, void *b, long m, long a, char *s);
void	myclocker_int(t_myclocker *x, long value);

// globals
static t_class	*s_myclocker_class = NULL;

/************************************************************************************/

int T_EXPORT main(void)
{
	t_class	*c = class_new("myclocker",
							(method)myclocker_new,
							(method)myclocker_free,
							sizeof(t_myclocker),
							(method)NULL,
							A_GIMME,
							0);

	common_symbols_init();

	class_addmethod(c, (method)myclocker_int,		"int",			A_LONG,	0);
	class_addmethod(c, (method)myclocker_assist,	"assist",		A_CANT, 0);
	class_addmethod(c, (method)stdinletinfo,	"inletinfo",	A_CANT, 0);

	class_register(_sym_box, c);
	s_myclocker_class = c;

	return 0;
}

/************************************************************************************/
// Object Creation Method

void *myclocker_new(t_symbol *s, long argc, t_atom *argv)
{
	t_myclocker	*x;

	x = (t_myclocker*)object_alloc(s_myclocker_class);
	if (x) {
		x->c_outlet = outlet_new(x, NULL);
	}
	return(x);
}

void myclocker_free(t_myclocker *x)
{
	if(x->thread_running > 0)
		systhread_join(x->thread, NULL);// avoid deadlock if thread should somehow be dead.
}

/************************************************************************************/
// Methods bound to input/inlets

void myclocker_assist(t_myclocker *x, void *b, long msg, long arg, char *dst)
{
	if (msg==1)
		strcpy(dst, "input");
	else if (msg==2)
		strcpy(dst, "output");
}

void myclocker_run(t_myclocker *x){
	long ctn = 0;
	while(1){
		post("hello hell... %ld", ctn);
		systhread_sleep(100);
		ctn++;
		if(x->thread_running == 0){
			systhread_exit(0);
			return;
		}
	}

}

void myclocker_int(t_myclocker *x, long value)
{
	// only trigger a change if values are unequal
	if (x->thread_running != value) {
		if(value == 1) {
			x->thread_running = 1;
			systhread_create((method)myclocker_run, x, 0, 0, 0, &x->thread);
		} else {
			x->thread_running = 0;
		}
	}
}


Just a quick comment, but it's hard for anyone to give you informed feedback without and example patch and object. 

However, from the above it looks like you are creating a new thread often for every incoming number (that doesn't match the last). Creating threads is more expensive than reusing them. And even still for something which is only using 3% CPU multithreading probably isn't going to give you a noticeable improvement, as multithreading has some CPU costs in thread management and synchronization.

I realize that this is just a minimal example you're using to learn the API, but if you're going to test efficiency, you should test with computation heavy work.

There's also some more advanced use of a pool of threads in the SDK "threadpool" example, and if you search the dev forums on "parallel" you'll find info on using the ext_parallel.h API, but these might be overkill for your purposes.

This is a very interesting topic and because many applications done with max are time-critical due to the focus on realtime capability.
I know that there are many thread programming tutorials out there but I would be interested in some more documented examples with max externals. Especially when it comes to the decision when to create threads and how to organise them.


Timothy's instructions are still helpful today (Max7, Windows)! Even without having the Tap.Tools tap.svn source code available, I was able to use his methodology almost exactly in my external and it worked like a charm.

>Here are the relevant bits:
>
>1. In the struct:
>t_systhread	thread;
>short	thread_running;
>
>2. In the free routine, you need to make sure that you don’t hang Max:
>if(x->thread_running > 0)
>systhread_join(x->thread, NULL);// avoid deadlock if thread should
>somehow be dead.
>
>3. Down in maxsvn_anything() we our flag in the struct and call a
>method, which executes in a new thread (whose handle we are storing in
>our instance):
>x->thread_running = 1;
>systhread_create((method)maxsvn_doanything, x, 0, 0, 0, &x->thread);
>
>4. When maxsvn_doanything() is done, and thus we don’t need the thread
>anymore, we do this:
>x->thread_running = 0;
>systhread_exit(0);

I second Kraftwerk's emotion on having a tutorial in the "Threading" part of the API documentation! Even if it's just Timothy's above instructions with example code. Reading through ext_systhread.h is good, but it's much easier to work from an example!


multi-threading in externals

innerrevolt

My brain is not capable of understanding the most common things within matrices. So, sorry for the stupid question. I literally have tried all day.

First of all, I am displaying audio through jit.catch. with a Matrix feedback. Simple right? Well, I wanted to change the color, and I can not wrap my brain around it.

I've gone through the tutorial 5 on Jitter ARGB, I've tried jit.brcosa (I think it only works with char), jit.scalebias, jit.fill, jit.op, jit.pack and unpack and I am just going crazy. I know it is a very simple thing to do, but I think I am just too new to Jitter.

How do I change the color from the cells that are being altered by the audio?

Anyways, If someone could help me, that would be awesome!

First of all, I am displaying audio through jit.catch.  with a Matrix feedback. Simple right? Well, I wanted to change the color, and I can not wrap my brain around it.

jitter-most-simple-question

your difficulties are probably to do with converting a float32 single plane to a 4 plane char-- this is one way to do it:

it uses jit.pack to duplicate the float 32 plane and jit,op to alter values in each colour plane-- note you need to trigger the output of the jit.pack with a bang, due to A-plane being first.


here's a hacked version which alters colour with audio amplitude:


I have so much to learn. Thank you so much!

I see that on that you normalized the jit.op, but the jit.normalize is not patched into anything. Is there an specific reason for this?

I love the hacked version with ZL and amplitude.

glad it's of value-- re the jit.normalize, i just used that to test a component of your patch, to see if it was working as intended, sort of a "debugging" tool, so you can just delete it


Jitter most simple question

Will

I'm trying to implement polyphony on an FM synth but I don't know why it isn't working. Been scratching my head and bouncing around tutorials on the internet all day trying to work out where my fault is. Could anyone help me?

I'm trying to implement polyphony on an FM synth but I don't know why it isn't working. Been scratching my head and bouncing around tutorials on the internet  all day trying to work out where my fault is. Could anyone help me?


poly-problems-4

Describe your problem in much more detail please. Says the teacher.


when I play one note, it was coming out fine and was being voiced properly. I thought I had everything working fine but then when I tried putting it into a poly object, I wasn't getting any sound output. I made sure to incorporate thispoly~ into the master/carrier asdr~ so that it would recognise note-offs and work properly etc but I'm struggling to diagnose the problem fully as I haven't been working with max that long.


Poly~ problems

Tobias Welby

I feel like there is probably a very simple solution to this, but at the moment i am stumped.

I need a bang to turn of a toggle once two separate variables have been met.

The input for both of these variable is different, so I have had difficulty using the Sel object, or the if statement.

Does anybody know if there is a way to do this?

I feel like there is probably a very simple solution to this, but at the moment i am stumped. 

I need a bang to turn of a toggle once two separate variables have been met. 

The input for both of these variable is different, so I have had difficulty using the Sel object, or the if statement. 

1-bang-once-two-input-criteria-have-been-met

[if ... ] sounds ok for this task.
See the attached patch.


Hi
[expr $i1==$i2] if both variables must match; [expr $i1==30 && $i2==80] if they differ. Then do [sel 1] for a bang

All of the above solutions make an assumption about which looked-for number will arrive first. Here's a more order-neutral method.

All of the above solutions make an assumption about which looked-for number will arrive first.

Nope. The first example is a [bondo] in front of an [if]
;-)


message ordering is NOT explicit to the novice; let them work it out for themselves? But Chris highlights this.

I refer the honourable gentleman to the comments I gave some months/years before


1 bang once two input criteria have been met?

daddymax

Ive been making a Waldorf Blofeld editor for ages, and have some time now to finish it, if i can get my head around a sudden left - turn.

So far, ive been getting the synth to send out sysex to Max - i twist the knob i want, record the string, make the pertinent section a changeable variable, and stick a knob on it - easy.

Now I get to the Arp section, and it seems that instead of mapping like this:

-----------------------------------------------------------------------------

1 16..112 128'..1/2' Osc 1 Octave, in steps of 12

327 0..127 0sssgaaa Arp Pattern Step/Glide/Accent 1

328 0..127 sss=0: normal Arp Pattern Step/Glide/Accent 2

Cycling '74 Forums

We are excited to announce the availability of our Eventide package for RNBO as a public beta. We’ve worked closely with Eventide to...

We are happy to announce the release of Max 8.6.2. This update focuses on Max for Live stability, fixing regression bugs, and a few new...

p.s. note that the doepfer endless dial is discontinued long ago, while the behringer comes with modified firmware and various user presets...

Nikolas, Very much appreciate the explanation on how to invoke the subdict and proper formatting! Strangely though, 2 problems came up...

Generaly peaking, you can probably do almost everything, but unless you give a bit more info, can't be sure. I assume you probably mean to...

the max-devkit uses git submodules to wrangle in max-api -- make sure you source both projects when attempting the walkthrough. if you...

hm - jweb does not work either - it does correctly load the webpage - but not the plugin - so i can't see the camera.

Timothy's instructions are still helpful today (Max7, Windows)! Even without having the Tap.Tools tap.svn source code available, I was able...

glad it's of value-- re the jit.normalize, i just used that to test a component of your patch, to see if it was working as intended, sort...

when I play one note, it was coming out fine and was being voiced properly. I thought I had everything working fine but then when I tried...

message ordering is NOT explicit to the novice; let them work it out for themselves? But Chris highlights this. favourite object is still...

Here's another hint.