🚀 go-pugleaf

In article <1992Sep24.140404.22842@cube.handheld.com>  
jmd@cube.handheld.com (Jim De Arras) writes:
> I was not able to attend the dev conf.  Does this mean we must have
> and believe 'inside' info, from one slide, over the public
> announcements?  Given that, how can we beleive anything NeXT says?
> It's all "marketing hype" in the end!

If I were you, I'd believe an engineer's inside info 10 times more often  
than I believed any marketing info.  This goes for any company, not just  
NeXT.  The nice thing about NeXT is that the marketing line is a lot  
closer to the truth than at other large computer companies.

The talk at NeXTWorld Expo was given about 6 months before 3.0 shipped.   
Back then, we knew enough about 3.0's performance to give developers a  
good idea of what components of the system were better, and which were   
worse.  This information is critical for deciding which tradeoffs to  
make in an application.

The exact contents of that slide were:

	3.0 Performance

	What's faster:
		Obj-C messaging overhead
		Precompiled headers
		Swaptimizer
	What's not:
		Overall performance

From an insiders perspective, let me just say that 3.0 is not perfect.   
We realize that -- we expect a lot out of our software too.  When it  
comes to deciding when to release a large software product like 3.0,  
there are a lot of tradeoffs to consider, and the final decision is  
always a compromise.

So, if you want to criticize 3.0's performance, keep in mind what was  
gained at the cost of slower software.  Keep in mind what NeXT's  
customers got for 3.0, when they got it, how much it costs, and what its  
overall quality is compared to other large software products.  Above  
all, keep in mind who NeXT's typical customer is, and what they want  
from their software.

Finally, let me say that constructive, well informed suggestions are  
listened to here at NeXT.  There's a lot we can do to improve  
performance in 3.0, and we'd love to hear your ideas.

Brian Pinkerton
NeXT Software Engineering