On Tue, January 6, 2009 08:45, Fred A. Miller wrote:
Mads Martin Joergensen wrote:
* James Knott <james.knott@rogers.com> [Jan 05. 2009 19:22]:
Some software may work perfectly on x32 while barfing on x64, especially software that wasn't designed with x64 in mind. Making software compatible with every conceivable architecture upgrade of the distant future would add massive layers of abstraction. Many years ago, while taking a C programming course, I ran into some "gotchas" with variable sizes, when moving between 32 & 16 bit systems. In class, we were using Borland's Turbo C++ for DOS and at home I had Borland C++ for OS/2. There was a difference in the size of some variable types, such as INT, so that a program that ran fine at home would fail in class, because I had exceeded the maximum integer size. However, I have no idea how 32 & 64 bit systems compare with this issue.
It's still the exact same thing. People are not making architecture independent code, still.
Supposedly, MickySoft is getting with the program and doing so. I have NO data to prove it, only from what I heard from the "inside." Anyway, IMHO, it's ignorant not to be writing 64-bit independent code. Make the effort now and increase performance. From now forward, there won't be anymore new 32-bit processors from Intel. It's a 64-bit world and has been for awhile.
It's not a mortal sin to write architecture dependent code, but only if you know what you are doing. For example, if you are coding in the embedded world on a 5MHz 8088 CPU with 32KiB RAM (I kid you not), it is ignorant to be writing hardware indepentent code. You won't be coding in Java, you will be coding in Assembler or C. In most other situations, please code architecture independent. I only want to make an argument against the usage of absolute words like "ALWAYS" and "NEVER". The world isn't black&white, it's shades of RGB. -- Amedee -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org