[Bf-committers] point cache

[Daniel Genrich]

Hello,

The problem of automatically and manually is the following: The system 
never knows what the users wants.
That's why I have "protect cache" and "Autoprotect cache" (if the user 
simulates many frames and forgets to press the "protect cache button", 
the "protect cache" button gets toggled).

The system can only guess if he's changing settings on purpose or an 
accident. Some user reported to me: Damn, 3 hours simulation gone, 
because I forgot to press "protect cache" and I was only "tab-tab"-ing. 
On the other hand, users also complain "why is cache always protected?" 
after I introduced the "autoprotect" feature. Of course that's right, 
because in the modelling/starting phase, you change settings alot and a 
protected cache only slows down the workflow. That's why I introduced a 
button to deactivate the autoprotect feature...How boring this all is :-D

The result: We can't do it right for all people but we should let behave 
all simulations equal! :-)

Daniel




Brecht Van Lommel schrieb:
>  However, I think we could do better automatically clearing the cache on user interaction. Right now it happens very often that you need to go and clear the cache because the system was not updated when settings or an effector changed. I thought the intention of the point cache was to have this working automatic, or should we give up that idea?
>