Re: following up on speed

[Daniel Weinreb <address@hidden>]

shivers@cc.gatech.edu wrote:

>
>GC is fundamentally an extremely global, non-modular thing -- it gets its
>fingers into every corner of the system. It has to, in order to understand
>who is using what. The payoff for this is modularity for the client software.
>
Actually, now that you put it this way, I realize that this is a really 
excellent example of the "Right Thing"
versus "Worse Is Better" dichotomy that Richard Gabriel discusses in his 
paper, the paper that was
the basis of the panel session at the LL1 workshop that I was on.  The 
dialogue reported in his paper, between
the "MIT Guy" (me) and the "New Jersey Guy" (Bill Joy), was about 
PCLSRing, but actually this GC
issue is a better example of a place where it is good to pay for a lot 
of complexity down in the underlying
system in order to provide cleanliness and simplicity for the user programs.

-- Dan