From: muta...@gmail.com   
      
   On Sunday, October 31, 2021 at 11:04:37 AM UTC+11, Scott Lurndal wrote:   
      
   > >> I'm now thinking that my pseudo-BIOS which exports fread etc   
   > >> to the OS, should be using the MVS flavor of the C library, which   
   > >> is focused on an expectation that the underlying system will   
   > >> give you a block, and you will be told the size of that block, and   
   > >> the underlying system does not provide a facility to keep track of   
   > >> a byte offset within that block for you.   
   > >>   
   > >> As opposed to a C library that can make use of a seek() syscall   
   > >> that will position on byte offset 3 or whatever in a file for you.   
   > >>   
   > >> I think that is the concept I have been missing.   
   > >   
   > >Is there existing terminology for this? Block devices versus   
   > >character streams perhaps?   
      
   > Congratulations, you've re-invented Unix 'block' and 'character'   
   > devices.   
      
   An ordinary file is considered to be a "device", is it?   
      
   Regardless, would you expect a C library to have different paths   
   for dealing with block devices vs character devices?   
      
   And does MVS deal exclusively with block devices?   
      
   When I went to implement PDPCLIB for MVS, I already had   
   a DOS/OS2 version with only minor differences, but for MVS   
   I knew I needed to deal with "records" and I wasn't sure whether   
   it could fit in with the code I already had, so I simply put some   
   #ifdefs and created brand new versions of fread/fgets/etc to   
   deal with MVS records.   
      
   It never occurred to me that the MVS flavor would be relevant   
   to Unix block devices.   
      
   BFN. Paul.   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)