Hello Mike,

On Jun 02 19:47 96, Mike Bilow of 1:323/107 wrote:

 MB> You are misinterpreting these test results.  You are 
 MB> allocating 1-byte allocations and forcing these to be 
 MB> committed, and you are seeing that OS/2 returns 4 KB blocks 
 MB> of committed memory on 64 KB address boundaries.  But the 
 MB> only entries being made in system tables are for the 4 KB 

but theseus/2 returns the memory descriptor table entrys. and Theseus/2
says size of block==64K. 

 MB> If a street has 10 houses and I choose to number them 100, 200, 300,... 
 MB> then the number of houses does not change from how many there would be if 
 MB> they were numbered more conventionally 1, 2, 3,...  It is absurd to worry 
 MB> about the supply of numbers instead of the supply of houses.

and if for some reason the highst number have 3 digits, you can only have 9
houses numbered in this street :-(((

and if I number the houses 1,2,3,...999 i can number 999 houses in the
street. thats the difference. if these is a bounded set, then there is
definityl NO injextiv and surjetiv function between the 2 numbering
methods.

(of course, if the sets are unbounded, there is a function. f(i)=100*i and
this function is injectiv and surjectiv)

(so, 1,2,3,... and 100,200,300,... have the same number of elements in an
unbounded set). 

and, as we all new, the memory is a bounded quantity. it must be <=
512MB. at a lot is preallocated with shared system resources. 

PS: i forgot that there is a limited number of descriptor table entrys (imo
something around 2**16) anyway. 

 WARPED!, Mario

--- Msgedsq/2 2.2e