35 lines
		
	
	
		
			1.0 KiB
		
	
	
	
		
			Plaintext
		
	
	
	
	
	
			
		
		
	
	
			35 lines
		
	
	
		
			1.0 KiB
		
	
	
	
		
			Plaintext
		
	
	
	
	
	
| 
 | |
| Close connection at halt (any change of at_halt/1 Vitor ?!!).
 | |
| 
 | |
| Investigate why plot_cpu/1 needs so much stack. (does it stilL ?)
 | |
| Investigate the x.y notation and see if SWI can cope with it.
 | |
| Calling functions natively ? 
 | |
| 
 | |
| We could drop the <- operator even, just hook an exception 
 | |
|    handler for predicates that are not there !!!
 | |
| 
 | |
| --- threads (SWI, specific ?)
 | |
| The R OS lib is thread-unsafe, add a Prolog layer to cope with this ?
 | |
| 
 | |
| Jan on Mailing list : 2013/1/10,
 | |
| 
 | |
| worker thread.  Just create a thread named 'R' and initialize R.
 | |
| 
 | |
| If you want to do something with R, use
 | |
| 
 | |
|     thread_self(Me),
 | |
|     thread_send_message('R', process(Me, <data>)),
 | |
|     thread_get_message(r_reply(Result))
 | |
| 
 | |
| and in the R thread, in a loop:
 | |
| 
 | |
|     thread_get_message(process(Sender, Data)),
 | |
|     r_compute(Data, Result),
 | |
|     thread_send_message(Sender, r_reply(Result)).
 | |
| 
 | |
| Of course, you need to make all this a bit cleaner and handle errors,
 | |
| etc, but this is the basic idea.
 | |
| 
 | |
| --- r_session compatibility add-on ? 
 | |
| one that would also solve the threading issue ?
 |