Re: Climber bug

[David Philippi]

On Sunday 27 October 2002 16:31, Gervase Lam wrote:
> 'Cos I thought that it was possible that the patch would not be committed
> as it didn't work perfectly.  Also, the change was quite simple.  If it
> was more complicated, I would have sent another diff.

If you ever show a piece of nontrivial, perfectly working code I'll call you a 
liar. ;-)
Your first patch went in the right direction for sure so why not apply it? If 
one can solve parts of a problem it may get obvious where the rest is for 
someone else. That you now provide the complete patch is the best way. ;-)

> I've been brought up on the "leave it alone 'cos it might break if you
> don't" principle.  Not that I do this all of time, of course...

One of the great advantages of CVS is that one can easily get back older code 
when something gets broken. I've broken Pingus code more then once but as 
long as I learn from my errors I won't let them stop me from making new 
ones. ;-)
I consider a CVS version that gets broken from time to time way better then a 
stagnating project where everyone tests for decades before commiting any 
change.

> In other words, sometimes the sprite was drawn BEFORE the direction of the
> sprite was updated.  Attached is the patch for this.

I got interested in why this did happen. The reason is simply that the 
fall/wall_actions of a pingus aren't updated. This is quite logical if one 
thinks about it but it does also mean that for persistent actions update() 
isn't called regulary through their lifetime. Since it's a bit hard to see 
this in the code unless one looks for it specifically we should probably 
either document or change this.
(I would not have noticed this if not for the incomplete patch. *g*)

Bye David