T.R | Title | User | Personal Name | Date | Lines |
---|
6082.1 | | DECWET::VOBA | | Thu Jun 05 1997 13:43 | 7 |
| Re .0, hmmm... that's not true. I've done it all the time.
However, there is an NTFS incompatibility between NT releases (starting
at SP2, i believe) that prevents earlier NT systems (regardless of
platform architecture) from seeing the disk properly.
--svb
|
6082.2 | | DECWET::SCHREIBER | DECeNT | Thu Jun 05 1997 14:18 | 11 |
| Incorrect except for a couple of specific instances:
1) Floating point data differences between x86 and Alpha architectures.
Binary data written on one architecture is not portable to the other.
2) If the disk is not dismounted cleanly (ie the system crashes),
moving the disk to a system of the other architecture is problematic
due to differences in the page size used to write the log file (page
size = memory page size, 4KB on x86 and 8KB on Alpha).
Benn
|
6082.3 | | DECWET::SCHREIBER | DECeNT | Thu Jun 05 1997 15:02 | 7 |
| re .2: IEEE floating point data is fully compatible. The issue only
exists for Intel 80-bit FP data, which shouldn't be a big problem for
most folks.
Sorry for the confusion on this.
Benn
|
6082.4 | | DECCXL::OUELLETTE | blackflies upgraded to mosquitos | Thu Jun 05 1997 16:16 | 4 |
| The 80-bit extend precision format is available from C/C++ using
the long double type on x86. Using it isn't such a good idea.
Most Fortran compiler for x86 don't allow REAL*10, including
Visual Fortran.
|