Re: gcc/informix 4gl running solaris getting segmentation faults

In article <200B76C1B6BDF308.DEBC56980334F5D3.540BE27767557AE2@library- proxy.airnews.net>, Daniel Wright <dw420@airmail.net> writes >Could be a number of things. > I would first try it using a different compiler, although I suspect >this is not an option for you. Then I would make sure everything >(including any forms that are referenced) are re-compiled and exist in >accessible directories (Strangely enough, this has caused core-dumps for >me, even though I believe a forms error should have been generated.) > >A couple of other things which have produced core dumps: > 1) Returning an integer to a function expecting a smallint. > 2) Accessing arrays out of bounds (They actually very often seem >to work, especially in 4GL, but don't count on it.....To test this I >wrote a test program which kept incrementing an array index in a loop. >Even though the array only had 5 elements, the program executed properly >through several hundred (750 or so, if I recall) iterations. A >variation of this test-case demonstrated that even before the core was >generated, the results were not reliable. > >You should probably give more information, though I suspect when you >investigate further, you will find the cause of your core dump.....Does >it always fail at exactly the same place? If it works when running >through the debugger (or w/ RDS), it is probably a variable >initialization problem. > >jpf wrote: > >> After compiling a 4ge with gcc compiler, getting core dump >> during excecution of program. Does any one have any idea what the >> problem >> is? > > Recompile with -fwriteable-strings This is in the 4gl release notes on some versions. > >-- >- Danny Wright >danwright@bigfoot.com >#include <stddisclaimers.h> >* God is real, unless declared as an integer. >* Why do programmers get Halloween and Christmas mixed up? > Because 0CT(31) == DEC(25) >* I'd love to change the world, but I can't get into the source code. >* /nev/dull > > -- David Williams