Re: LYNX-DEV Configure patch for SCO

[T.E.Dickey]

> Jean-Pierre Radley wrote:
> 
> > SCO Unix (both 3.2v4.x and 3.2v5.x) needs to inlude libsocket when 
> > compiling.
> 
> As Tom said, this should be found by configure (and always has been, for
> me, but I'm still on ac.36).  Special cases should be reduced, not
> increased.  For which I say:
right - the special cases are mainly those for which I've insufficient
information to design a reliable test.
 
> >   sco3.2v5*)
> >     test $ac_cv_prog_gcc != yes && CC="cc -belf"
> > +   LIBS="$LIBS -lsocket"
> >     ;;
> >   sco*)
> > !   LIBS="$LIBS -lmalloc -lsocket"
> 
> In fact, the "-lmalloc" should also be eliminated as a special case.
> configure could check whether a libmalloc exists, and use it if it does
> (unless there are known cases of libmalloc being worse than the libc
> malloc, on some OS?)
I don't think that I have access to any machines with a -lmalloc at the
moment (that's just taken from Makefile.old, to cover whatever configuration
it was originally added for).
 
> Even the "-belf" could be configured.  The gcc's for OpenServer
> variously support either "-belf" or "-melf" for ELF code generation.
> ELF is generally preferable.  I'm sure configure could figure out which
> of those flags does anything useful...  "-melf" should be more general,
> available under gcc on multiple OSes and architectures, so a generic
> test for it might be sensible.
I see only one mention of "-melf" in the info for gcc (perhaps it is an
obsolete or rarely used option).
 
> >     # It's a little late to do much, but try to warn user if he's using
> >     # SCO's cc (which is reported to have broken const/volatile).
> 
> If there are problems that affect Lynx, they can be configure-tested...
I suppose so - if I know what the symptoms are (the comment was based
on email from you, where you recommended using either rcc or gcc because
of problems with const/volatile).  I've only built with cc, so I'm not
sure what problems there might be (at least it does _compile_ and _run_,
though there may indeed be runtime problems that I don't encounter with
my short tests).
 
> >Bela<

-- 
Thomas E. Dickey
address@hidden
http://www.clark.net/pub/dickey
;
; To UNSUBSCRIBE:  Send a mail message to address@hidden
;                  with "unsubscribe lynx-dev" (without the
;                  quotation marks) on a line by itself.
;