[PEAK] dispatch questions

[Phillip J. Eby]

At 06:34 PM 3/14/05 -0800, Dave Peticolas wrote:
>I realize that when('x != 5') would be a lousy way of specifying the
>default case, but I don't understand why the two outputs would differ
>for the non-default case (x == 5). Also, it now seems that the order
>doesn't change when I use the latest checkout of PEAK. Is this a bug
>that has been fixed or is the order of those two 'before' methods
>undefined?

I recently fixed a bug in strategy.method_list() that had the effect of 
making certain method combinations usable only once.  I don't know whether 
it might also have had the effect you're describing, but I can tell you 
that "before" and "after" methods use strategy.method_list() for 
combining.  So, there's a chance that fixing the one bug also fixed the other.

Aside from this, I should note that the ordering for before and after 
methods that are of the same specificity is not well-defined anyway, and 
might change in the future.  Currently, it's the same as definition order 
for "before" methods and reverse definition order for "after" methods, but 
I think perhaps it should be the other way around.  It seems to me that 
methods defined later should be "more before" and "more after" than methods 
defined earlier.