> I noticed something like this problem while building X11R6.1 under
> NetBSD/hp300 and NetBSD/sun3.  As far as I can tell, this is specific
> to the m68k ports, and it seemed to appear when we upgraded to
> gcc 2.7.2.

This is probably due to the way that gcc/m68k elects to access symbols
in shared libraries. I don't know for sure, but it might be the case
that the function address is stored in the Global offset table (this is
not the case for, eg. the sparc and i386), which implies that ld.so
wants to resolve those address at startup time, so you don't have the
`lazy evaluation' of function calls.


> I must admit that I don't know where to start looking to solve the
> problem... In the mean time, it may be worth also implcitly including
> libtermcap with libg++.  This is a simple Makefile trick,

I also agree with jtc that removing all references to curses from libg++
would be a desirable thing.

-pk