11OLSEN

How do I get data to Max from a function that has been called without passing *x (the struct pointer).

Is it legal to have a global variable holding my x pointer so I can access the object struct (mutex,clocks,data)? 

It works this way but are there any risks or leaks with this method ?

Hi.
How do I get data to Max from a function that has been called without passing *x (the struct pointer).
Is it legal to have a global variable holding my x pointer so I can access the object struct (mutex,clocks,data)? 
It works this way but are there any risks or leaks with this method ?
O.


Hi.
How do I get data to Max from a function that has been called without passing *x (the struct pointer).
Is it legal to have a global vari

how-to-reach-my-struct-without-passing-it-to-a-function

Ok, one problem becomes obvious. A second instance of the object will put the struct pointer to the same global variable.

I'm trying again. You know sometimes there are libs or OS related features that implement a callback function. That means something happens and a predefined function in your code gets called. So there is no way to pass the obligatory *x to these function. How do i find a way back to my object from there?

All correctly implemented callback functions will allow the user to pass a void* pointer to the function.

All callbacks in the Max/MSP SDK do, the Max/MSP perform routine does and I can't remember a callback function from a different API that doesn't.

If you have a concrete example please bring it up.

I would be surprised if a professional/commercial API had a callback function implemented in a way that doesn't allow user access.

All correctly implemented callback functions will allow the user to pass a void* pointer to the function.
All callbacks in the Max/MSP SDK do, the Max/MSP perform routine does and I can't remember a callback function from a different API  that doesn't.
If you have a concrete example please bring it up.
I would be surprised if a professional/commercial API had a callback function implemented in a way that doesn't allow user access.

for example hooking the mouse in Windows. At some point you register the hook 

mouse_event_hhook = SetWindowsHookEx(WH_MOUSE_LL, mouse_hook_event_proc, hInst, 0);

which is a function called by the same thread that registred the hook

LRESULT CALLBACK mouse_hook_event_proc(int nCode, WPARAM wParam, LPARAM lParam) {

    MSLLHOOKSTRUCT *mshook = (MSLLHOOKSTRUCT *) lParam;

Hi,
for example hooking the mouse in Windows. At some point you register the hook 
mouse_event_hhook = SetWindowsHookEx(WH_MOUSE_LL, mouse_hook_event_proc, hInst, 0);

which is a function called by the same thread that registred the hook
LRESULT CALLBACK mouse_hook_event_proc(int nCode, WPARAM wParam, LPARAM lParam) {
	MSLLHOOKSTRUCT *mshook = (MSLLHOOKSTRUCT *) lParam;
	switch (wParam) {
            case WM_LBUTTONDOWN:  ... no x here...

Good question and example. I think your not missing anything.

Reading your previous posts I now get the whole picture... your only alternative is to use a global variable, however if you use a global then - when you have more than one instance in a patch - you will get a conflict. Clear.

The choice of not allowing the user to pass a custom pointer to the SetWindowsHookEx() function is partially explained because you are expected to have just one instance of a given hook. It's a very low-level API call, so it makes sense.

You need to deal with the limitation that there can be only one hook running, no matter how many instances of your object the user creates in a patch. The only function that gets called only once when you instantiate a Max external is the main() function. So a possible approach would be to bring some of the hook setup code into the main() function. In the new method of every object instance check if there is a hook already created and running. If not, create it and refer to it. If yes, refer to the one already created. This means you could potentially have many instances of your object refer to the same hook, which brings into the picture other potential threading issues, however I believe that's your only option.

I know it's a lot more easily said than done...

Good question and example. I think your not missing anything.
Reading your previous posts I now get the whole picture... your only alternative is to use a global variable, however if you use a global then - when you have more than one instance in a patch - you will get a conflict. Clear.

The choice of not allowing the user to pass a custom pointer to the SetWindowsHookEx() function is partially explained because you are expected to have just one instance of a given hook. It's a very low-level API call, so it makes sense.

I know it's a lot more easily said than done...
Sorry I could not be of more help.

Thanks Luigi. I should have been more specific from the first post on. 

At least now I know it only works with a special solution.

And yes that brings up a lot of new questions ;)

Thanks Luigi. I should have been more specific from the first post on. 
At least now I know it only works with a special solution.
And yes that brings up a lot of new questions ;)


How to reach my struct without passing it to a function?