core dumps with 4gl program

We switched to another development server for Y2K reasons, some Informix-4gl programs compiled on the new box core dump at run time. The old server had Informix-4gl version 4.11, the only media we could find for the new is Informix-4gl version 4.12, the O/S is SunOS 4.1.3_U1, the compiler is an old version of gcc. Hard to believe, huh? We did put our applications through a battery of Y2K related tests. This is the output, it coredumps when it tries to initialize a variable defined as money(20,2) to 0.00: after assigning the nitems Memory fault This is the 4gl code, Gsubtotal is defined as money(20,2): display "after assigning the nitems" let Gsubtotal = 0.00 This is the esql code that gets generated: $ display "after assigning the nitems"; status = efcode; if (status < 0) { fgl_fatal(fgl_modname, 489, status); } pushstrdec(" 0.00 "); popdec(&gsubtotal); status = (chkprec(&gsubtotal, 5122)) ? -1226 : 0; $ display "after assigning the subtotal"; Any ideas? Perhaps an object that wasn't recompiled? We have millions of lines of legacy code, code and libraries in perhaps 100-200 different locations on the server (I didn't write this crap), most everything has been recompiled, but it's possible I missed something. Thanks, Alan

Something in a library somewhere not recompiled would be my guess. Art S. Kagel Alan wrote: > > We switched to another development server for Y2K reasons, some Informix-4gl > programs compiled on the new box core dump at run time. > > The old server had Informix-4gl version 4.11, the only media we could find > for the new is Informix-4gl version 4.12, the O/S is SunOS 4.1.3_U1, the > compiler is an old version of gcc. Hard to believe, huh? We did put our > applications through a battery of Y2K related tests. > > This is the output, it coredumps when it tries to initialize a variable > defined as money(20,2) to 0.00: > after assigning the nitems > Memory fault > > This is the 4gl code, Gsubtotal is defined as money(20,2): > display "after assigning the nitems" > let Gsubtotal = 0.00 > > This is the esql code that gets generated: > $ display "after assigning the nitems"; > status = efcode; > if (status < 0) > { > fgl_fatal(fgl_modname, 489, status); > } > pushstrdec(" 0.00 "); > popdec(&gsubtotal); > status = (chkprec(&gsubtotal, 5122)) ? -1226 : 0; > $ display "after assigning the subtotal";> > Any ideas? Perhaps an object that wasn't recompiled? We have millions of > lines of legacy code, code and libraries in perhaps 100-200 different > locations on the server (I didn't write this crap), most everything has been > recompiled, but it's possible I missed something. > > Thanks, > Alan

Alan wrote: > We switched to another development server for Y2K reasons, some Informix-4gl > programs compiled on the new box core dump at run time. > > The old server had Informix-4gl version 4.11, the only media we could find > for the new is Informix-4gl version 4.12, the O/S is SunOS 4.1.3_U1, the > compiler is an old version of gcc. Hard to believe, huh? We did put our > applications through a battery of Y2K related tests. > GCC is the giveaway -- unless you use -fwriteable-strings, you will get core dumps. SunOS 4.1.3 is pretty damn ancient. I4GL 4.12 is pretty ancient. You need to consider upgrading a lot. Do not modify SCCS files after the 31 December 1999; you will corrupt your files. You are definitely treading on the thinnest of wafer-thin ice. > > This is the output, it coredumps when it tries to initialize a variable > defined as money(20,2) to 0.00: > after assigning the nitems > Memory fault > > This is the 4gl code, Gsubtotal is defined as money(20,2): > display "after assigning the nitems" > let Gsubtotal = 0.00 > > This is the esql code that gets generated: > $ display "after assigning the nitems"; > status = efcode; > if (status < 0) > { > fgl_fatal(fgl_modname, 489, status); > } > pushstrdec(" 0.00 "); > popdec(&gsubtotal); > status = (chkprec(&gsubtotal, 5122)) ? -1226 : 0; > $ display "after assigning the subtotal";> > Any ideas? Perhaps an object that wasn't recompiled? We have millions of > lines of legacy code, code and libraries in perhaps 100-200 different > locations on the server (I didn't write this crap), most everything has been > recompiled, but it's possible I missed something. > > Thanks, > Alan -- Jonathan Leffler (jleffler@informix.com, jleffler@earthlink.net) Guardian of DBD::Informix v0.62 -- see http://www.perl.com/CPAN #include <disclaimer.h>