memory

[Fergus Henderson]

On 04-Feb-1998, Peter Schachte <pets at students.cs.mu.OZ.AU> wrote:
> Here's my proposal:
> 
> 	Word MR_make_permanent(Word term, Word type_info);
> 
> 	Returns a copy of term that can safely be stored in C's memory
> 	somewhere even if it may be accessed after failure.  Of
> 	course, type_info is a type info structure for term.
> 
> 
> 	Word MR_make_partially_permanent(Word term, Word type_info, 
> 					 Word *limit);
> 
> 	This is the same as MR_make_permanent, except that if limit is an
> 	address on the heap, parts of term that are "older" than limit will
> 	not be copied.  This is useful when you know that the permanent copy
> 	of term will not be accessed after the heap pointer has backtracked
> 	beyond limit.  Naturally, this always occurs when the permanent term
> 	is to be stored in *limit.
> 
> 	I'd like to describe the limit argument without referring to the
> 	"heap," but don't see how to.
> 
> I believe these two functions, as part of the documented standard Mercury
> runtime, should be enough.  In conservative GC grades, these functions just
> return term, otherwise they allocate the "global heap" zone if it hasn't
> already been allocated, and copy term there, returning a pointer.  The
> resulting pointer could be used as if it were an ordinary heap term.  Of
> course, the global heap is never reclaimed so garbage collection is pretty
> important, but I'll leave that for Tyson to worry about.  What else is there
> to say?  I guess the size of the global heap zone should be handled like the
> solutions heap zone. 
> 
> Comments?  Have I missed something?

That looks OK to me.

-- 
Fergus Henderson <fjh at cs.mu.oz.au>   |  "I have always known that the pursuit
WWW: <http://www.cs.mu.oz.au/~fjh>   |  of excellence is a lethal habit"
PGP: finger fjh at 128.250.37.3         |     -- the last words of T. S. Garp.