Together with the help from the c4dlounge.eu forum the issues is solved.
The main issue was I did not read the manual good enough.

"Note that Size symbolizes the axis size and not the expansion of the object itself. For planar projections, the true dimension is the two/fold of the axis length."

I did some more testing and I think I got it now.
In MouseInput() get and set your coordinates.
Then call c4d.DrawViews(). This will call Draw(), where you can (and should) do the drawing like bd.DrawLine2D().

class PlaceTool(c4d.plugins.ToolData):       	

    v1 = c4d.Vector(0,0,0)
    v2 = c4d.Vector(0,0,0)
    
    def Draw(self, doc, data, bd, bh, bt, flags):	
        bd.SetPen(c4d.Vector(256,256,256))
        bd.DrawLine2D(self.v1, self.v2)
        return True
        
    def MouseInput(self, doc, data, bd, win, msg):
        self.v1 = c4d.Vector(0,0,0)
        self.v2 = c4d.Vector(msg[c4d.BFM_INPUT_X], msg[c4d.BFM_INPUT_Y], 0)
        c4d.DrawViews(c4d.DRAWFLAGS_ONLY_ACTIVE_VIEW|c4d.DRAWFLAGS_NO_THREAD|c4d.DRAWFLAGS_NO_REDUCTION|c4d.DRAWFLAGS_STATICBREAK)  #Added 
        return True
        
        

Thanks, great explanation!
One small issue. Delete doesn't work because objParent' is not defined.

Traceback (most recent call last):
File "scriptmanager", line 251, in DeletePressed
NameError: global name 'objParent' is not defined

Here the code that, I think, solves the issue:

    def DeletePressed(self, root, userdata):
        "Called when a delete event is received."
        for tex in reversed(list(TextureObjectIterator(self.listOfTexture))):
            if tex.IsSelected:
                objParent = tex.GetParent()               # Added
                listToRemove = objParent.GetChildren() if objParent is not None else self.listOfTexture
                listToRemove.remove(tex)

@heilei said in Best plugin type for background (thread) processing?:

Py-TextureBaker

That is a very good working example.
I rebuild my plugin using py-texturebaker and now it is working!

After some studying, I think @PluginStudent is fully correct when referring to the global scope.

@PluginStudent said in Best plugin type for background (thread) processing?:

Please read my post above and the C4DThread Manual carefully.

You can start the thread in a CommandData plugin. But this plugin must not own the thread instance. The thread instance should be stored in the global scope.

In the MessageData plugin, you should react to a core message sent from the thread. Nothing else; no creation or waiting.

I guess it is working because in the dialog the thread class is initiated.
textureBakerThread = None
and further in the code it is referenced with self.textureBakerThread.

I was using

        thread = UserThread()
        thread.Start()

Everybody thanks for all the help.

@PluginStudent said in Reading script file into memory:

You have to define the APIS in the projectdefinition.txt of your plugin.

Not in the projectdefinition.txt of the solution.

Great, tht solved it. Thanks!

To summarise, there are 2 projectdefinition.txt files.

1. on main / sdk level, Define which project (plugins, etc.)
   Example:

Platform=Win64;OSX
Type=Solution
 
// defines the level of rigour of the source processor's style check
stylecheck.level=0

Solution=\
	plugins/cinema4dsdk;\
	plugins/maxonsdk.module;\
	plugins/commandline;\
    plugins/CommandLineRendering

2. on plugin level, // Configuration of a custom solution in the projectdefinition.txt file
   Example:

// Supported platforms
Platform=Win64;OSX

// Type of project
Type=DLL

// this plugin depends on these frameworks:
APIS=\
  cinema.framework; \
  misc.framework; \
  image.framework; \
  core.framework; \
  python.framework

// Enable some advanced classic API support; not needed for hybrid plugins
C4D=true

// Plug-in code-style check level
stylecheck.level=0

// Custom ID
ModuleId=net.maxonexample.commandlinerender

Yes, I am referring to that page.
I saw that statement, but I assumed it was the main definition for the python plugin.

  // set __name__ = __main__
  scope.Add("__name__"_s, maxon::Data("__main__"_s)) iferr_return;

If I understand you correctly, I can add different parameters with different values.
For example:

  // set __param1__ = "c:/temp"
  scope.Add("param1"_s, maxon::Data("c:/temp"_s)) iferr_return;
  scope.Add("param2"_s, maxon::Data(220)) iferr_return;

Yes, I must read the manuals more throroughly (RTFMS).
Excuses: it is so much and I just wanted to compiler R20 in R21.
But you are correct.

BUT, adding iferr to the code worked, AND also solved the other issue.
Everything is working now!

Thanks!

Great, that works!

One other question.
I have now a script that removes old takes and adds takes based on selected objects

Now I see that the main take still has 'old' previous deleted takes.
I do not think that is an issue, but I am not sure.

I want to create Takes using Python
I have a Null with an object as a child under it.

Below code creates the Take, but it does not fill the Category ok.

What am I doing wrong?

import c4d

def main():
    # Gets the TakeData from the active document (holds all information about Takes)
    takeData = doc.GetTakeData()
    if takeData is None:
        raise RuntimeError("Failed to retrieve the take data.")
   
    obj = doc.GetFirstObject()
    print ("First: ", obj.GetName())    
        
    # Adds a new Take
    newTake = takeData.AddTake(obj.GetName(), None, None)
    if newTake is None:
        raise RuntimeError("Failed to add a new take.")

    # Defines the created Take as the active one
    takeData.SetCurrentTake(newTake)

    c4d.CallCommand(431000081) # Auto Take toggle On
    # Unhide object, initially object is hidden.
    obj[c4d.ID_BASEOBJECT_VISIBILITY_EDITOR] = c4d.OBJECT_ON     
    obj[c4d.ID_BASEOBJECT_VISIBILITY_RENDER] = c4d.OBJECT_ON 

    c4d.CallCommand(431000081) # Auto Take toggle Off
    # Hide object again
    obj[c4d.ID_BASEOBJECT_VISIBILITY_EDITOR] = c4d.OBJECT_OFF      
    obj[c4d.ID_BASEOBJECT_VISIBILITY_RENDER] = c4d.OBJECT_OFF
        
    # Pushes an update event to Cinema 4D
    c4d.EventAdd()


if __name__ == '__main__':
    main()

Ok, clear. Thank you.

Ok, did some test and what I see is that DrawMsg() is called every time the vertical scroll bar is clicked.
My understanding was that cinema handles this internally.

But because DrawMsg() is called I draw all thumbnails and thus not taking into account where the vertical scroll bar is located. So if only the lower 25% must be (is) shown, I redraw all thumbnails.

Can I calculate which part of the UA must be drawn (where is vertical scroll bar and how large is it)?

I do understand that without some code it is very hard to answer.
I will try to put only the code, which I think causes the bad performance, in a test plugin.

Also, you mention "You could maybe parallelize this in C++" and "just draw everything on a GeClipMap". Do you have some examples for these options?
Also, I do not understand the option "to draw only visible bitmap". All the thumbnails I draw are rectangles (500x500).

I have a folder with a lot of thumbnails.
I am displaying the thumbnails in a UA with a vertical scrollbar.

Vertical scrolling is all handled by cinema 4d, I have added no code for scrolling.
So far so good. For a small number of thumbnails it is working ok, but for more thumbnails the performance is very bad!

I noticed, after adding some debug statements, that the bad performance seems to be outside DrawMsg().
DrawMsg() itself seems fast enough, but before anything happens on the screen it takes some time.
Also when moving the vertical slider, DrawMsg() is called more than once.

What can I do to increase performance?
Build my own scrolling functionality?

Here the basics of DrawMsg()

    def DrawMsg(self, x1, y1, x2, y2, msg):
      
        self.OffScreenOn() 
        self.SetClippingRegion(x1,y1,x2,y2)                    

        self.DrawSetPen(c4d.Vector(.4))    
        self.DrawSetTextCol(c4d.Vector(1,1,1), c4d.Vector(.4))   #GeUserArea.DrawSetTextCol(fg, bg)
        self.DrawRectangle(x1, y1, x2, y2)  # Clear UA by Draws the UA rectangle area

        # display thumbnails
        ...
        for rows in range(0, self.nrRows):    
            for colums in range(0, self.nrColumns):  
                ....
                self.DrawBitmap(self.bmp, x1, y1, self.thumbnailSize, self.thumbnailSize, 0, 0, 500, 500, c4d.BMP_NORMAL)
            
        return

Example of the UA,

Ok, I found out that the issue is in "c4d_symbols.h"
There I defined FILE = 1010011.
After removing the FILE definition, it was ok again.
Setting FILE = 10002 did not solve the issue.

So removing it and defining FLE somewhere else, will be ok.

enum
{
ResBasedMenu =  101001,
MY_CHECKBOX  =  101002,
CURR_SIZE_STR = 101003,
NEW_X =         101004,
NEW_Y =         101005,
NEW_Z =         101006,
MY_TEXTBOX =    101007,
BUTTON_CANCEL = 101008,
BUTTON_OK =     101009,
BUTTON_RESET =  1010010,
// FILE =          1010011,
// FILE =          10002,
MYBITMAPBUTTON =1010012,
IDC_CUSTOM1 = 1010013,
}

I installed (R23) our plugin (MeshBoolean), but now the File entry of the Content Browser gives a "StrNotFound".
In R20 it was ok.
The console shows no warnings or errors.
Choosing another layout or another Node Space does not help.

But you can still select the 'File' option.

@m_magalhaes said in Different behavior on Mac for Commanddata options:

hi,

thanks for linking the other thread and yes, that's the difference with the mac menu on the osx bar.
On those menu we can't add the cogwheel so we choose to display the option by default (with was the old behaviors)

cheers,
Manuel

Ok, clear.
Please mention it in the documentation.

-Pim

What I also notice, is that - on the mac - when I do a customize command and add the plugin as an icon to my interface, the option icon is there and it works like expected.