[gclist] Fast-allocating non-copying GC's

[David Ungar]

Allow me to paraphrase for brevity:

David Chase makes the points that free list allocation is so fast
that the allocation overhead does not prevent him from using new objects,
that the problem of fragmentation does not strike,
and that the a relocating collector
has to make
"other tradeoffs with respect to
system design and/or optimization and/or compatibility with uncooperative
code."

David, I would believe that these assertions could obtain for C and C++,
but would be surprised if they were true for Smalltalk and related languages.
In which arena does your experience lie?


Also the statement: "I'm in this for the engineering, not the religion."
is hard to understand.  Are you trying to denigrate a specific assertion
by casting it as religion? Which one?




-- Dave