On Thu, Apr 16, 2009 at 5:28 PM, Adam Lowry <[email protected]> wrote: > On Apr 15, 2009, at 5:25 PM, kirby urner wrote:
<< SNIP >> > Registration was the only case I could think of that worked well in my mind. > In the past I've got the augmentation route and felt that it made the code > less readable and less understandable; I should have composed a helper > object. In this example that method or a straight subclass (or even mixin) > would be easier to read in my opinion. > Thank you. I appreciated your outburst at the meeting, take your warnings about reduced readability seriously (your Python scholarship has always impressed me). We went down a long road on edu-sig awhile back, using function decorators to turn a function into its mathematical derivative. I'd even showcased this technique in a talk (GIS meeting in 2005, similar to the one next week). Later in the thread, in which Guido participated, we reached a consensus that this was hurting more than helping, more a "false trail" than a great way to go. ** So yeah, I'm happy to keep a retreat route and think your suggestion to hunt class decorators in the wild is a good one. Kirby > But for pure library code like an ORM, where the user is not intended to > need to dive into the guts, they can be useful. > > Perhaps we can find some code in the wild that make good use of them? > > Adam > _______________________________________________ > Portland mailing list > [email protected] > http://mail.python.org/mailman/listinfo/portland > ** there's a running game expats play, involving "hares" laying a trail ahead of time, others following, with a number of dead ends deliberately included, just to confuse, make it more like a maze: http://www.gthhh.com/ (my dad was super into it, in both Bangladesh and Lesotho, mom too some, and me when I'd visit). _______________________________________________ Portland mailing list [email protected] http://mail.python.org/mailman/listinfo/portland
