RedHat 6.0 and float data

I have an annoying problem on RedHat 6.0. I've got the eval set from Intraware (RDS 7.20 UD7, SE 7.24 UC5 and IDS 7.3 UC7) and am trying to port my 4GL apps from SCO to the new platform. The problem seems to be some sort of storage anomaly. My programs do very strange things with float data types, both in program manipulations, and insert/selects from the database tables. An example; let floatvariable = 0 display floatvariable --> 0.') insert 0 to table select it back and display --> error 'data does fit in float variable' If I convert fields and variables to decimal, then this problem goes away. Informix support have the ball, but I thought I'd try here as well, in case it's an 'unofficial' known problem. Bryan Tonnet batonnet@phase4.com.au

I wrote: > I have an annoying problem on RedHat 6.0. I've got the eval set from > Intraware (RDS 7.20 UD7, SE 7.24 UC5 and IDS 7.3 UC7) and am trying to > port my 4GL apps from SCO to the new platform. > > The problem seems to be some sort of storage anomaly. My programs > do very strange things with float data types, both in program > manipulations, and insert/selects from the database tables. An > example; > > let floatvariable = 0 > display floatvariable --> 0.') > > insert 0 to table > select it back and display --> error 'data does fit in float variable' > > If I convert fields and variables to decimal, then this problem goes > away. Informix support have the ball, but I thought I'd try here as > well, in case it's an 'unofficial' known problem. > > Further to this, I installed RH6 on a completely different machine using vanilla Intel hardware. I've tried both SE and IDS on both machines and not only do I get the above listed errors, but also; Some variables (some of the time) do not change value after a LET statement DISPLAY some_var --> 0.') LET some_var = 1 DISPLAY some_var --> 0.') Some variables (some of the time) are initialised, or contain rubbish data after a FOREACH loop starts. LET some_var = 1 DISPLAY some_var --> 1 FOREACH some_curs DISPLAY some_var --> 3.234e-32 END FOREACH The problems now do not appear restricted to float types, I am having symptoms from integer types as well. The data in the tables looks fine, problems occur once it hits an RDS program variable (some of them, some of the time....) The problems are pretty much reproducable on both platforms, so I no longer suspect any weirdo hardware issues. Is anyone running the above combination of O/S, RDS and Engine and having any success? Informix support claimed 'no problem' today, but they did their testing on RH5.2. Unfortunately, I can't use this version for hardware reasons. TIA Bryan Tonnet

I'm not sure if this will help, but I'm running redhat 6.0, 4gl 7.20ud7 and SE7.24uc5. I compiled the examples as shown and had no problems. I did have problems running code at first, check my previous post on redhat 6.0, 4ge segmentation fault. Perhaps it's restricted to RDS. I don't have that installed yet but will let you know the results once I do. I'm also running on an upgraded 5.2 to 6.0 system, not a fresh install. ray

In <375E67D0.E61A8A0D@erols.com> ray beckler <rbeckler@erols.com> writes: >I did have problems running code at first, check my previous post on >redhat 6.0, 4ge segmentation fault. Yep, I read your thread with interest the other day, but didn't act on it. My belief is that fglpc doesn't go outside of INFORMIXDIR and the compat libraries won't useful in this instance. If I'm wrong (wouldn't be the first time) I'll surely make that my next step. Hmmm, how do you persuade RDS to look at a different library set? >Perhaps it's restricted to RDS. I don't have that installed yet but >will let you know the results once I do. Great! It would be terrific to get another's opinion. I can provide code, schema and data if you want it for testing (ie your own stuff works fine). Bryan Tonnet batonnet@phase4.com.au -- Bryan Tonnet batonnet@phase4.com.au