🚀 go-pugleaf

In article <MLM.95Sep23000627@cslab3e.cs.brown.edu>, mlm@cs.brown.edu wrote:

> Not quite true.  UNIX programmers assume malloc/free will do a good
> job of reusing memory that has been free()d, so they don't have to
> worry about it.

No.  They assumed that their memory allocation patterns combined with a
fairly large backing store make malloc/free a good enough solution that
fragmentation isn't a practical problem.  It isn't like malloc/free do
anything magical.

>If they neglect to free their memory, they lose - but

Different problem.  What if the heap is so fragmented that there isn't a
large enough contiguous space of memory?  There are three solutions: 
abort, compact the heap, or have the programmer then go and figure out how
to make a big enough contiguous free block.  Only the first two are
practical, and as a user, I prefer the second method.
-- 
        Nevin ":-)" Liber       nevin@CS.Arizona.EDU    (520) 293-2799