Christian. It is probably best to split this question into 5 differnt posts. Some people might be hesitant to answer if they don't know the answer to all five. Try a really specific title to each post.

>Hello all,
>
>I know this is long but if I don't solve this problem a client's 3 year investment into a system will not be able to be installed at any of his client sites. Furthermore, 3 of his clients are unable to bill at all due to Y2K problems w/their old systems and I'm over 1000 miles from home installing those sites now. Imagine what it will be like if next week with a new system they still cannot bill. Obviously this is critical.
>
>The problem:
>
>I have a process that generates billing transactions. It works fine for smaller billings, i.e. when billing companies with a small number of members. When I bill a company with about 1400 members (possibly less but certainly w/1400) at some point the data environment gets fried. This is exhibited as follows:
>
>1) C0000005 error; or
>
>2) "Cannot redefine ." error
> occurs at a line:
>
> m.dPmtDate = DATE()
>
> but this can occur at any variable defined at this point.
> I've even inserted a command:
>
> LOCAL m.myVar1, m.myVar2 (etc)...
>
> and get the same cannot redefine error ON THE "LOCAL" command instead
> of the subsequent "m.dPmtDate = DATE()" command.
>
> One time when I suspended and held the cursor over the m.dPmtDate
> variable in the debugger, the tooltip text looked like this:
> "DATE()??"
> I do not actually know which binary characters were displayed
>
>3) "Alias '' not found" error. When this happens and I suspend execution I can simply try to browse a table from the data session window and will be unable to browse it, just get the same "Alias '' not found" error. When this error occurs I can get the following crap in the debugger:
>
> ALIAS() "myalias"
> USED('myAlias') .T.
> RECNO('myAlias') 550
> myAlias.myField (Expression Could Not Be Evaluated)
> myField "John"
> FIELD(1) "myfield"
>
> As you can see, the environment is totally fried
>
>4) "Property does not exist" When trying to call a method that does exist. In the debugger:
> PEMSTATUS(omyObject,'myMethod',5) .T.
>
>In this case I can (while suspended) go to the command window and instanciate the object and attempt to call a method and get the same error. PEMSTATUS returns .T. but executing the method from the command line gives error "Property does not exist"
>
>
>Areas of suspicion:
>
>The major process that always completes before the symptoms occur is a SCAN with a large number of processes w/in the loop. These processes include several different SQL SELECTs, and BEGIN/END TRANSACTIONs (not nested transactions though). When the number of iterations of the scan is low, the environment stays healthy and subsequent processes finish. When the number of iterations is 1400 (possibly less but certainly at 1400) then subsequent processes fail as outlined above.
>
>Attempted solutions:
>
>I have put FLUSH in various parts of the process
>
>When suspended I have CLOSE DATA ALL and FLUSHed and still the "property does not exist" problem occurs with new objects.
>
>Set the MVCOUNT=60000 in the config.fpw.
>
>Logged SYS(1001) and SYS(1016) values to text files during the process up to the C5 error. PLENTY of memory left available to VFP.
>
>I have created objects as PRIVATE outside of reiterative processes rather than letting the reiterative processes repeatedly create and release the objects.
>
>I have changed some called methods with SQL SELECTs to methods that open tables under other aliases and SEEK records, etc. rather than using a SQL, however this is not a practical solution to implement universally.
>
>I have attempted to explicitly .NULL. local objects in methods rather than letting the method complete and release the local objects themselves, but have not done this universally yet.
>
>I have disabled norton antivirus auto-protect.
>
>
>The final rub:
>
>The process ALWAYS fails under Win98, Fails about 50% of the time under NT Server 4.0, but has never failed yet under NT Workstation 4.0. We have tested all of the following with the EXACT SAME DATASET (although failures have been documented with different datasets). We restore the beginning data every time before testing.
>
>-Consistent failures on three different Win98 machines.
>
>-Consistent failure on one NT Server and 50% failure on another NT Server (both backoffice SP4). The successes under NT Server were with Norton AV Auto Protect disabled, but it has also failed with it disabled. We have never succeeded with it enabled but this could be conincidental.
>
>-100% Success on two different NT Workstations (SP5). One w/Norton NAV installed and running and the other w/out NAV at all
>
>We have tested under NT Workstation with a billing as high as 3300 members, however one of the clients we are installing now has a billing run of over 20,000 members. I'm worried that our successes under NT Workstation will not continue under such higher billing runs because the failures seem to be member-quantity related.
>
>I greatly appreciate any suggestions for further testing or problem resolution. Thank you so much for taking your valuable time to assist me in this issue.
>
>
>
>Christian Berrigan
>Spartacus Development
>mailto:spartacs@e-z.net