@dan rathbun said:

The API docs say this won't work... because the file is already open (on PC,) by the time the AppObserver gets loaded.

Dan,
In my script:
When double-click on SU file

1. opening the model in SU
2. execute PRODUCE.sceneobserver (automatically)
3. A Appobserver is also created (which fires PRODUCE.sceneobserver again, but since there already is an observer there is no new one created)
4. When a model is then opened from the file menu the Appobserver is called

Logically and after testing this seems to work ok.

My code here is not complete though, since in my code there is another check to see if the framobserver must be created or not (depending if the opened model has some specific settings)

Dan,

I have done some testing with this and it seems to work yes.

Thx for cleaning up the script!

For me it's simple:

For registered users give the opportunity to comment on incorrect information/give more explanation and give possibility to add code examples. I always like the way PHP does it: http://php.net/manual/en/function.include.php

If you give us the tools, we help you with the content...

We needed a Sceneobserver being attached whenever a model is openend (through the menu item or when double clicking a skp file in Windows Explorer.
And did it like this:

Ath the end of the ruby file:

PRODUCE.sceneobserver
# Observer that watches the application for opening of models from the file menu and creates scenes observer automatically.
     class MyAppObserver < Sketchup;;AppObserver
       def onOpenModel(model)
			PRODUCE.sceneobserver
   end
end

This handles both possibilities.
Now to avoid creating 2 observers just check if there is already an observer present

def PRODUCE.sceneobserver #add the scene observer after PRODUCE.scenes or when scenes need to be reloaded
	#
	if ($xDframesobserverid==nil or $xDframesobserverid=='') #no frame observer existing or it has been removed
	#UI.messagebox ('make observer')
		$xDframesobserverid=Sketchup;;Pages.add_frame_change_observer(XdFrameObserver.new) 
	end
end

Hope this helps

'explains code to an inanimate object'
I'd prefer this:

Downside is that after a while you'll need to debug what you are saying

great, thank you both for the information!

ok, hitting 'Expand' would be a good idea

I use this code to list all used component definition names:

def TESTXD.listcompnames1
		$componentnames=[]
		Sketchup.active_model.entities.each{|e|
			TESTXD.listcompnames2(e)
		}
		$componentnames.uniq!
		$componentnames.sort!
		puts ($componentnames)
	end
	def TESTXD.listcompnames2(e)
		theclass=e.class
		if theclass==Sketchup;;Group 
			gen=e.entities
			gen.each{|gent|
				TESTXD.listcompnames2(gent)
			}
		elsif theclass==Sketchup;;ComponentInstance
			$componentnames<<e.definition.name
			e.definition.entities.each{|ent|
				TESTXD.listcompnames2(ent)
			}
		end
	end

This gives me a list in the ruby console.
Now if I compare this list with the list shown in the components windows I see that certain definitions are available in ruby but not in the components window.

Is this due to the fact that those component definitions are only used in nested component definitions? That makes no sense.
Or maybe I'm missing something.

Excellent information.
Thank you ThomThom, much appreciated!

@thomthom said:

What went wrong? Examples?

It goes wrong caused by the fact that the entities object is edited while executing the loop. So as you stated.

@thomthom said:

No, just when you modify it. If you just read states and properties there should be no problems.

Modify (like in push pull) or only when the entity is deleted?

@thomthom said:

The second - using .erase_entities.

In the second case I could use Sketchup.active_model.entities.each (without the to_a) Correct? And if so would that be faster then with to_a?

Thx again!