Re: [clisp-list] Successful compile of Clisp - but the memory fault/coredump ...

[Vadim K. <va...@gm...>]

Frank Goenninger wrote:
> Dear Clispers,
>
> being completely new to Clisp (but not to Lisp) I am trying to get  
> Clisp running on HP-UX 11i (HP-UX 11.11) on PA-RISC1.1-based machine.  
> I get a compile done, but when the Makefile calls
>
> ./lisp,run ...
>
> for the first run I get a memory fault with resulting coredump.
>   

your problem resembles mine when I compiled using MSVC++ in Windors.
The fix came into the CVS, but the #ifdef-ed compilers aren't fixed, 
namely HP-UX one.

Look for the following code in lispbibl.d (this is after fixing)

# Unspecified length of arrays in structures:
# struct { ...; ...; type x[unspecified]; }
# Instead of sizeof(..) you'll always have to use offsetof(..,x).
#if defined(GNU) || defined(MICROSOFT) /* GNU & MS C are able to work 
with arrays of length 0 */
  #define unspecified 0
#elif 0
  # Usually one would omit the array's limit
  #define unspecified
#else
  # However, HP-UX- and IRIX-compilers will only work with this:
  #define unspecified 1
#endif


Even with comment about HP-UX compilers, it do not justify wrong 
aligment logic.
try modifying this file according to your situation and see how it goes.

> Any attempt to get more info from Clisp failed - I assume the  
> coredump is happening quite early after startup...
>
> Any idea how I can debug this?
>
> Thx for any pointers. Any help really appreciated.
>
> Regards
>      Frank
>   

Best regards,
Vadim.