From: muta...@gmail.com   
      
   On Friday, July 9, 2021 at 11:44:39 PM UTC+10, muta...@gmail.com wrote:   
      
   > But this introduces another issue. Callers have   
   > traditionally cared about being properly aligned, but   
   > segmentation introduces a requirement for the   
   > caller to be able to request that the entire block   
   > be within a single segment. Huge memory model   
   > callers don't care about that, but everyone else   
   > does, so some sort of indicator, or different   
   > INT 21H function code would seem to be   
   > appropriate.   
      
   I think we have the following use cases:   
      
   1. Less than 64k memory requested, caller expecting   
   an offset of 0.   
      
   2. More than 64k memory requested, caller expecting   
   an offset of 0.   
      
   3. Less than 64k memory requested, caller can handle   
   a non-zero offset, but the memory block must be fully   
   contained within the segment returned.   
      
   4. Less than 64k memory requested, caller can handle   
   a block spanning a segment, but the offset must be 0.   
      
   5. More than 64k memory requested, caller can handle   
   both a non-zero offset and the memory spanning more   
   than 1 segment.   
      
   I think implementing 4 should be delayed until there is   
   an actual use case though.   
      
   5 is what huge memory model programs should be requesting,   
   to be nice.   
      
   3 is useful for non-huge memory model programs when   
   segment shifts are increased to 16 bits or whatever.   
      
   1 and 2 are the only thing MSDOS provides, and are   
   distinguished by the number of pages requested.   
      
   BFN. Paul.   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)