Re: 4GL application core dumps

I spoke too soon on this one. We did find a problem in the PREPARE statements. When preparing at statement from a character string, the value of the string gets expanded, ie. a blank space is placed around each parenthesis and comma, thus expanding the length of the character string, thus writing to memory outside the boundary of the string, thus causing a sgementation fault and core dump. Sample program showing the error: main define str char(2000), i smallint LET str = "CREATE TABLE aaaa1920", " (tmpline SMALLINT,", "fcode SMALLINT,", "action CHAR(1),", "invnum CHAR(16),", "upflag CHAR(1))" # #-----display value of str and its length # display str clipped let i = length(str) display "length = ",i # #-----prepare string str # prepare test_stmt from str # #-----display value of str again and its length # Note: value of str is changed!! and it's length # display str clipped let i = length(str) display "length = ",i end main Output from program on Solaris 5.7, I4GL 7.30.UC1, and IDS 9.20.UC1: CREATE TABLE aaaa1920 (tmpline SMALLINT,fcode SMALLINT,action CHAR(1),invnum CHAR(16),upflag CHAR(1))length = 101 CREATE TABLE aaaa1920 ( tmpline SMALLINT , fcode SMALLINT , action CHAR ( 1 ) , invnum CHAR ( 16 ) , upflag CHAR ( 1 ) )length = 123 We're passing the info along to tech support to get fixed. Candy -- Candy McCall ONLINE Computing, Inc. (candym@olcinc.com) >Well, it looks as if it's just the order of >installation of tools and engines. Our >customer re-installed and thus far >no core dumps. > >Candy-- >Candy McCall >ONLINE Computing, Inc. >(candym@olcinc.com) > > > >>We have a customer who upgraded their 4GL and >>engine: >> >>They were previously at: >> >>Solaris 5.6 >>IDS 7.23.UC2 >>I4GL 6.04.UC1 >> >>They upgraded to: >> >>Solaris 5.7 >>IDS 9.20.UC1 (Foundation 2000) >>I4GL 7.30.UC1 >> >>Now they are encountering core dumps >>in the 4GL application when previously >>they were not encountered. >> >>As anyone encountered any problems with >>this combination? We haven't as yet analyzed >>exactly where in the code the core dumps >>happens, but wondered if anyone else has >>encountered this problem. >> >>Thanks in advance for any help! >> >>Candy >>-- >>Candy McCall >>ONLINE Computing, Inc. >>(candym@olcinc.com) >> >