problem in a VC++ application compiled at NT run at Windows 2000 3

[lingliu]

A VC++ application is compiled at NT, when it runs at Windows 2000, it can not produce correct float type calculation results. But when it runs at NT , results are correct. I changed the structure alignment from 8byte to 1, 4,16 and no difference is found.

Would compiling at Windows 2000 solve the problem?

Thanks!

 

[Cagliostro]

maybe it is a problem with endian format. Ion Filipski

filipski@excite.com

 

[joeGrammar]

yes, yes it would fix the problem, remember, just because its windows doesn't mean its the same architecture.

 

[Zyrenthian]

Yes... compile it on 2k

Matt

 

[lingliu]

THank you all. Unfurnately, compile at 2K is not a fix. The results are the same. But if you want to a program run at 2K, you must( better?) to compile it at 2K?

 

[joeGrammar]

That may be a windows 2000 glitch. It may be worth reporting

 

[Cagliostro]

please send me the project if is not too big. Ion Filipski

filipski@excite.com

 

[Lim]

I compile my programs on NT and it runs fine on WIN2000. And they are pretty complicated. It can be just a bug that was hiden on Nt. Check memory corruption may be you have uninisialize memory read or read out of array boundary...

 

[lingliu]

The cause of this problem is found. When NT transfer a date time type to a string , it transfers it to DD/MM/YY , but
Windows 2000 transfers it to DD/MM/YYYY.

Thanks a lot and nice to talk to you!

 

[joeGrammar]

sweet candycakes