🚀 go-pugleaf

In article <CA7MpC.uC@cs.uiuc.edu>, morrison@cs.uiuc.edu (Vance Morrison) writes:
  matomira@di.epfl.ch (Fernando Mato Mira) writes:
 
  >Garbage collection to the Operating System!!
  >Trying to put together languages each one having their own GC run-time
  >is a mess and definitely wasteful of machine an development resources.
  >Given the current trend in gnu.misc.discuss concerning the GPL I propose
  >that we search to create a new organization: the Free Garbage Foundation!! :-| 

  I believe to some extent this is an impossible dream.  What a garbage 
  collector does for you is free you from dealing with resource allocation
  of a very common resource (memory).   The problem with garbage collection
  is that in its niave (current) form it is NOT modular.  That is most
  garbage collection techniques rely on a single monolithic algorithm that
  assumes that the entire memory system is under its control  While this
  can be a reasonable assumtion for memory on the scale of a single application, 
  it IS a STRONG assumption.   For example it is usually NOT a reasonable
  assumption for a fileing system (which is why we don't see many 'garbage
  collectors' for disks, even though such a system would be quite useful).  
  It is also the root of the problem with interfacing with foreign 
  languages (since code written in these languages are not under the
  garbage collectors control).  
   
  On a more fundamental note, garbage collection abstracts away the problem
  of resource (of memory at least).  When resource allocation is not the
  problem we are trying to solve (as is the case in most applications),
  this is a real plus, but in those cases where resource allocation IS
  the problem to be solved, it makes the problem even impossible to 
  express (at least not directly)!!  A major function of operating system 
  and many concurrent/networking algorithms is in fact resource allocation.
  Imposing a single standard is likely to be inappropriate.
  

Uk,uk. I don't know if what you got was that I would like to impose a GC on
all processes of a system, including the OS, but if that was the case, that's
not what I meant (after all, this is a Unix world, isn't it? [$$$-) <-- guess who]).
The "Memory manager" should be able to divide a virtual memory space into different
areas, according to management policy (eg. non-garbage collected, generation scavenging, mark&sweep, etc.),
and object representation formats (that is, a protocol should be provided to specify tagging schemes, 
tag-to-box-format tables, etc.). In that way the different applications can use the facility if and as
they want.

Regarding the FSF, such a facility could well be one of the "special" cases, libc style.
I guess you could get quite a few donations from C++ vendors. Even the LISP
vendors would profit from it. 


-- 
Fernando D. Mato Mira
Computer Graphics Lab			  "There is only one Language
Swiss Federal Institute of Technology	    and McCarthy is its prophet"
matomira@di.epfl.ch

FAX 	 : +41 (21) 693 - 5328

Disclaimer:

disclaim([],[]).
disclaim([H|T],[DH,DT]) :- fixed_point_disclaimer(H,DH),
			   disclaim(T,DT).
fixed_point_disclaimer(Text,fixed_point_disclaimer(Text,_)).