From: muta...@gmail.com   
      
   On Thursday, November 10, 2022 at 5:12:55 AM UTC+8, Joe Monk wrote:   
   > > No, just because my 32-bit executable can tolerate being run   
   > > on environments that offer less than 32 address lines, or more than   
   > > 32 address lines, or exactly 32 address lines, does not alter the   
   > > fact that it is a 32-bit executable (which is why it can only address   
   > > 4 GB of memory, even if you make infinite address lines available).   
   > >   
   > > Although it would still be amusing to find out what your actually   
   > > definition of "x-bit executable" is. You don't seem to be able to   
   > > identify an executable that cannot address more than 4 GB   
   > > as 32-bit. Every time IBM comes up with a new machine (or if   
   > > they have built multiple machines, which they have), you run   
   > > around in circles trying to figure out what number to give it.   
   > >   
   > > You don't seem to be alone. This mainframe "culture" is one of   
   > > the things that overcomplicates it to the point that people who   
   > > have been programming on it for decades end up running   
   > > around in circles when someone from an independent environment   
   > > points out their misconception.   
   > >   
   > It is YOU who doesnt understand. You use a 32-bit pointer. So what?   
      
   So - that limits the executable to addressing 4 GB of memory,   
   and classifies it as a 32-bit executable.   
      
   It means that the executable itself is inherently capable of using up   
   to 32 address lines, and no more.   
      
   You can give it less address lines, and it will address less memory, or   
   you can give it more address lines, and they will be unused.   
      
   What possible alternative definition could you have for the   
   term "32-bit executable"?   
      
   > If I take a 32-bit pointer and stick it in a 64-bit register, I am STILL   
   > addressing 64-bits. All 64-bits are significant.   
      
   Sure. And the OS, ie z/PDOS-generic, will deliberately make use   
   of those extra address lines, and e.g. store the number 1 or 2   
   in the high 32 bits,so that the 32-bit executable mentioned above   
   will happily run at the 4 GB or 8 GB mark.   
      
   But that is happening completely outside the (inherently)   
   32-bit executable's knowledge.   
      
   > And in AM64, the addresses arent limited to 2^32 -1.   
      
   I didn't say they were.   
      
   I said something quite different. I said that I produce 32-bit   
   clean executables (like IEFBR14 and not much else because   
   of a crap (in some ways - like they didn't use 3-way diffs either)   
   mainframe culture with a very poor understanding), which   
   (because they are clean) are inherently capable of running   
   in any AMODE - anything from 1 to infinity.   
      
   BFN. Paul.   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)