Re: 4gl return value problem

daisy wrote: > >"June C. Hunt" <june_c_hunt@hotmail.com> wrote in message >news:<2sd8c.52394$Fh4.329@twister.nyroc.rr.com>... > > daisy wrote: > > > I had a question occurred when my 4gl main program calling a > > > function. There seems to be something wrong with the return value > > > > > > Statements in my main program : > > > > > > ud_fun() > > > ........... > > > let p=1 > > > while p > > > if del_fun() then exit while end if > > > let p=0 > > > end while > > > if p then > > > rollback work > > > prompt "error... " for char answer > > > end if > > > ............ > > > end function > > > > > > > > > del_fun() > > > ........ > > > if m_ac31('2',n.acm02,n.acm00,n.acm01, > > > o[i].acn02 thru o[i].acn15) then > > > display "m_ac31 error ",o[i].acn02 at 21,1 > > > sleep 5 > > > return 1 > > > end if > > > .................. > > > end function > > > > > > statements in my function > > > > > > m_ac31() > > > ............... > > > if status or sqlca.sqlerrd[3]<1 then > > > display "Upd ac31 err-",status > > > sleep 5 > > > return 1 > > > end if > > > ...... > > > > > > > > > My problem is that I 'sometimes ' found that the m_ac31 > > > function was performed successfully (because there is no error message > > > displayed ) > > > but the results comed out wrong ; which means there were error > > > during the function was called but my main program did not catch the > > > return value .Could any one tell me where the problem is ? > > > > > > My Informix was old SE. I do not know what is the problem in my > > > program. > > > > Since you don't specify what you mean when you say the results are >wrong, > > I'm just guessing at where to start. Was data changed that shouldn't >have > > been? Was data skipped? Were one or more unexpected records changed? >Were > > one or more expected records missed while others were selected? Simply > > saying the results are "wrong" leaves a whole lot of possibilities. > > > > Without knowing exactly what "wrong" means, my first suggestion would be >to > > change the error checking in m_ac31. I would check SQLCA.SQLCODE rather > > than STATUS - or check both. The STATUS variable will indicate errors >from > > both the form and from SQL statements whereas SQLCA.SQLCODE indicates >errors > > (or success) of just the SQL statements. Since you only show the error > > checking, we can't be sure what comes before it in your code; I'd be >more > > comfortable checking the SQLCA record. I would also be inclined to >evaluate > > SQLCA.SQLERRD[3] against a known value. For example, I would find out >how > > many records I expect to process, and would compare SQLCA.SQLERRD[3] >against > > that value rather than zero. Is it possible that the problem isn't >that no > > records were processed but rather more records than you were expecting >were > > processed? With these changes, at least you'd be sure you were checking >the > > results of the last SQL statement and that you were processing as many > > records as you were expecting. Note that this doesn't necessarily mean >the > > correct records were processed, but at least the actual number of >records > > can be determined. > > >Sorry that I did not describe my question clearly. >The statemenst in m_ac31 were as follows > update ac31 set acp05=acp05-damt, > acp06=acp06-camt, > acp07=acp07-amt, > acp08=acp08-amt, > acp09=acp09-ramt > where acp00=x.acn00 and acp01=x.acn04 and>acp02=ym > and acp03=dno and acp04=x.acn05 > if status or sqlca.sqlerrd[3]<1 then > let damt=damt*-1 > let camt=camt*-1 > let amt=amt*-1 > let ramt=ramt*-1 > insert into ac31 values> >(x.acn00,x.acn04,ym,dno,x.acn05,damt,camt, > amt,amt,ramt) > if status or sqlca.sqlerrd[3]<1 then > display "Upd ac31 err-",status > sleep 5 > return 1 > end if > end if > > What bothered me was that sometimes I found that the record to be >updated were actually exist, but after the program were completed and >check the result >I found that the record were not updated at all and the return value >did not pass to the calling procedure (in my situation that's function >del_fun()). > > Once I manually cleard the wrong result and I run the program again, >but it >comed out right . I don't know if it's problem of file lock , or is >there any >mistakes in my program. > > Please give me some advice . > Thanks again No obvious mistakes in the program - what we can see of it, anyway. Is it possible that the UPDATE was rolled back as a part of a larger group of transactions in the ud_fun function? I'm also a little curious about the purpose of the WHILE loop around that function call, but for the moment will assume that you are showing just a portion of that code. What else is going on between the BEGIN WORK/COMMIT WORK in that function? If it were a locking issue, I'd exect it to be caught with the SQLCA.SQLERRD[3]<1 check (assuming that a single record is being identified by the WHERE clause and that multiple records aren't being updated). Even so, I prefer to lock the record myself before attempting an update - just to be sure that I've got it. So - without seeing the rest of the code, my best guess is that the transaction was rolled back. What else is going on in ud_fun? -- June Hunt _________________________________________________________________ All the action. All the drama. Get NCAA hoops coverage at MSN Sports by ESPN. http://msn.espn.go.com/index.html?partnersite=espn sending to informix-list