🚀 go-pugleaf

In article <444ulgINN1bp@keats.ugrad.cs.ubc.ca>, c2a192@ugrad.cs.ubc.ca
(Kazimir Kylheku) wrote:

> That's just the static code size. What about the total count of
> executed instructions?

Fairly minor.  In practice, you don't end up locking/unlocking the handles
all that often, unless you are a very naive programmer.

>I bet that goes up quite a bit more.

It all depends on what you mean by that vague term "quite a bit".  I'd
guess no more than a percent or two, looking at typical code.  You lose
far more in breaking your code into function calls, or waiting to page
something off of a disk.  If you wish to microoptimize, go ahead.

> time you have to dereference a handle, you have to execute twice
> as many instructions. Chasing dynamic links becomes slower.

> Passing the handle means you have to do an extra dereference.

Again, microoptimizing.

>  Knuth says
> it's crap in The Art of Computer Programming, though in more scholarly
> terms. This week, I'm going to dig up the precise quote.

Go ahead.  As I said before, you're really good at spewing out factoids at
us, but really bad at doing actual design and analysis of real systems.

> When I first found out that the Mac uses handle-based memory management, 
> I laughed so hard that my buttons just about flew off my shirt.

Pinheads tend to do that.  The rest of us would rather learn *why* they
implemented it the way they did.  That is why we look at books on the
subject at hand, such as Inside Macintosh, and *learn* about it.  All
you've done is read Knuth and gotten a bunch of factoids out of it; the
secret to knowledge is knowing what it applies to and what it doesn't, not
in regurgitating the information.  Good luck in your career as a game show
contestant on Jeopardy! or as a master of Trivial Pursuit.
-- 
        Nevin ":-)" Liber       nevin@CS.Arizona.EDU    (520) 293-2799