Re: 4GL Core Dumps. Help!

Path: emory!swrinde!cs.utexas.edu!sdd.hp.com!hplabs!nsc!pyramid!infmx!aland From: aland@informix.com (Colonel Panic) Newsgroups: comp.databases.informix Message-ID: <1991Sep24.045337.22349@informix.com> Date: 24 Sep 91 04:53:37 GMT References: <1991Sep23.141552.27704@cbfsb.att.com> Sender: news@informix.com (Usenet News) Organization: International Brotherhood of Geeks, Local 619 In article <1991Sep23.141552.27704@cbfsb.att.com> nll@cbnewsf.cb.att.com (neal.l.leitner) writes: > >I have run into a little bit of a problem using Informix 4gl. The problem >is Memory Fault - Core Dumps. Let me try and explain it a little: >We are using the Informix-SE and Informix 4GL running on an AT&T 6386/33 >with 8 MB of RAM. The paramaters are tuned for an eight meg machine running >Unix 3.2.3. Informix does not give any indication of any special paramater >settings, so we didn't change anything. What has happened is. A Form is >opened and displayed. The program then goes into an INPUT BY NAME function >and within the INPUT BY NAME there is: > > ON KEY (CONTROL-W) > CALL SHOWHELP(1) > >So, if you are on the first field of the form and hit control-w, it will >sometimes (note sometimes) dump a core. It will always dump the core until >you re-compile it and then it works fine!!! Weird? I'll say! I'll say, too. *Nobody* has reported such behavior in the past. There *is* a known problem with CONSTRUCT: if you enter query criteria in the last field of a CONSTRUCT set, it can core dump on certain platforms. This is bug 7628, fixed generically in 4.00.UG1 and in some port-level patch releases (for example, there is a 4.00.UD3 on your platform that has this fix). Of course, if recompiling "fixes" it, as you say, it sounds like something in your *system* is going haywire. Have you tried making backups of the "working" .4ge and cmp'ing it to the "failing" ones? Could the .4ge be getting clobbered? Or, something else with the system? That box has very thorough memory diagnostics (we have several here); have you tried running them any time recently? Ours have fried SIMMs three times in a year. >you re-compile it and then it works fine!!! Weird? I'll say! Of course, >the Informix "No-Help" Technical Support People are clueless and never return >any calls whatsoever. The assigned engineer entered in the system that he spoke to you on Thursday and you had said that you found that the problem did not recur when -lmalloc was specified and that you'd be out of the office until today. Is he lying? >Neal Leitner >...att!emdbl1!nll -- Alan Denney aland@informix.com {pyramid|uunet}!infmx!aland Disclaimer: These opinions are mine alone. If I am caught or killed, the secretary will disavow any knowledge of my actions. "I like getting on elevators when it's just me and a kid. Then, I push *all* the buttons, somebody else gets on, and he gets the blame!" -- Lorne Elliot