 | |  |  |  |  |
| | | | | |
The addresses of elements on a databases are now counted in bytes rather than in 'words':
| Version 6 | Version 7 | |
| Base size of element | Word | Byte |
| File Address Offset | Word | Byte |
| File Start Address | Word 1 | Byte 0 |
The C API now fully complies to the new
standard of version 7. However, for compatibility reasons, the 'old'
Fortran API functions still adhere to the
version 6 standard. Examples: The Fortran API
function calls prpdb and lgtdb
return the 'word' address, i.e the number of 32 bit elements, starting
at element 1. The 'word-addressable' paged file access functions,
like
subroutine rdvdb(handle,ipos,buffer,nword,status) subroutine wrvdb(handle,ipos,buffer,nword,status) subroutine wtwadb(handle,buffer,ipos,nword,status) subroutine rdwadb(handle,buffer,ipos,nword,status)
operate on 'word' address, i.e the number of 32 bit elements,
starting at element 1, and the number of elements
nword is counted in 'words', i.e in 32 bit
units.
In the new C API the type of some arguments automatically adapt to
32 or 64 bit platforms. Thus, mcSize,
mcOff are 32 bit or a 64 bit integers. Note that the
size information is always stored in 64 bits on the database!