Traits approaching mainstream Squeak
Blake
blake at kingdomrpg.com
Tue Aug 30 22:48:00 UTC 2005
I'm actually quite surprised by some of the discussion here. I'm not
surprised people are skeptical; this is a good thing. But I'm a little
surprised by the defense of inheritance and by things "being good enough
because they work now". I mean, to me, Smalltalk has always been about
ideals. Sure, things that work "well enough" should be at the bottom of
everyone's list of things to break, but the flaws in the collection
hierarchy are obvious--even if we've all managed to work around them (or
not conern ourselves with them).
As for defending inheritance, I learned about OO programming back in 1990.
A johnny-come-lately, I guess, since I'd been programming for ten years
and (in lieu of any formal training or knowledge of internet) struggling
to come up with systems of expressions in traditional languages that would
give me just a fraction of what OOP did. And yet, within weeks of learning
OO, the limitations of inheritance were apparent even to me. Yeah, you can
almost always insert a feature at the highest point in the hierarchy you
need it, but then you end up with big, fat objects stuffed with lots of
unused code. It's inelegant, it's wasteful and--in the case of
collections--it leads to code duplication. It also strikes me as a
security hazard.
Is any of this seriously debated? I mean, traits notwithstanding, has
anyone who has ever built a large hierarchy of objects NOT run into the
above problems?
===Blake===
More information about the Squeak-dev
mailing list
|