Vb.net on windows 98 is unstable?

[jamesrhaynes]

My application does precise mathematical calculations involving extracting roots with a QR algorithim. The test case is a difficult-to-extract matrix. The program works fine in both debug mode and in release mode when run under WinXP. Performance on Win 98 however is very odd. A debug version of the exe can run fine and give the correct results on Win98. A release version of the exe (with source unchanged) fails to calculate correctly. Sometimes, even a debug version of the exe will fail on Win 98 with just a minor change such as adding a comment or even recompiling with no change! The same code in VB6 runs perfectly on all Windows machines. This anomally has been verified on 4 different Win 98 machines with pentium ranging from I to IV and all exhibit this unstable behavior. Good operation has been verified on 2 XP machines and one Win 2000. Has anyone else discovered this unstable behavior of .Net on Win 98?

 

[kevinf2349]

I had a problem with the date being returned strange on a Win98 box. The date was always the same date in 1600! So I gave up trying to run .NET applications on that machine. Fortunately it is only my home PC so it isn't a big deal for me.

 

[chiph]

While Win98 is listed as a supported platform for .NET, many users here have reported odd problems. I myself wouldn't write for it.

Chip H.


____________________________________________________________________
If you want to get the best response to a question, please read FAQ222-2244 first

 

[chrissie1]

yep i had the same weird problems with win98 ( a certain combobox that freezes the application)

Christiaan Baes
Belgium

What a wonderfull world - Louis armstrong

 

[jamesrhaynes]

This is the thread originator reporting back with what we discovered and our workaround. Out application creates the matrix from a dynamic model description. In Win 98, when the matrix is first created by detailed calculations, the first time a NaN value is stored in each double precision cell rather than the correct number. This is even though the number itself is checked and is ok and well within range of double precision arithimitic. We created a (user choice) loop check the existence of NaN in the matrix and if so to repeat the whole process again and sometimes on the second or the third time it actually takes the double precision number and all goes well. Even so, it is not consistent. Sometimes it takes 2, 3 or even more times to get it to work. This anomaly never happens on Win XP or any other valic .net system and it always works correctly the first time. So, for the time being, that is our Win98 work-around to just ask for a "try again". This is not a great solution, but after hundreds of experiments, its the best we can do short of droping Win98 from the compatibility list.