dbus-1.0.2 patches for hpux & others

[Havoc Pennington]

Hi,

Thanks, I (or someone) will have to go through these and massage/apply 
these changes to the latest version. They all look roughly right, thanks 
for sending them.

My notes on what to massage:
  - I would comment the AC_SYS_LARGEFILE that we are relying on
    it eliminating the need for the _POSIX_PTHREAD_SEMANTICS
  - strtoll.c in tools/ doesn't make sense, need to figure out
    where it goes (also it's more dbus-normal to do this ourselves
    with an #ifdef HAVE_STRTOLL code block and a wrapper function, than
    with AC_REPLACE_FUNCS)
  - I have to look up what LIBOBJS is
  - in the vsnprintf-error-handling code, whitespace isn't right
  - on Linux, isn't 0 a legitimate vsnprintf return if the format
    string is ""? will need handling.
  - the 1024*640 max alloc size looks fishy - if we need this,
    something is wrong
  - would prefer a solution to the AIX poll.h problem other than
    putting a system header in sysdeps.h, or at least would prefer
    to only put that header there on AIX
  - to fix "inline" should "#define inline" on platforms that
    don't have it rather than removing from all platforms
  - dbus-shell.c relocation of string.h I don't understand

Havoc