Re: LYNX-DEV ac#54: FreeBSD bulding is broken again, please fix

[Андрей Чернов]

On Thu, 21 Aug 1997, T.E.Dickey wrote:

> > checking if we can link ncurses without -lmytinfo ... + echo 
> > configure:3819: checking if we can link ncurses without -lmytinfo
> > + cf_ncurses_SAVE=-lncurses -lmytinfo
> > + echo+ sed -lncurses -e -lmytinfo s/-lmytinfo //
> >  -e s/-lmytinfo$//
> > + q=-lncurses
> > + test -lncurses  != -lncurses -lmytinfo
> > + LIBS=-lncurses
> > ^^^^^^^^^^^^^^^^^^ (wrong)
> Something is missing from your log (I did run this script on FreeBSD 2.1.5
> the other night, to debug/test it, and it worked properly then, though I'm
> still mystified why you insist on using ncurses 1.8.6).  Here's the full

FreeBSD 2.1.5 is too obsolete to deal with. I care about FreeBSD 2.2.3 and
FreeBSD 3.0 -current.  There is no such thing like ncurses 1.8.6 for
FreeBSD 2.2.3 and 3.0 -current, there is merged 1.8.6+4.1 ncurses now. Raw
ncurses 4.1 is not imported yet, ask address@hidden if you interested
why, he deals with clean import. I prefer to make merged ncurses version
right now and commit it than wait for clean import forever.

>       AC_TRY_LINK([#include <$cf_cv_ncurses_header>],
>               [initscr()], 
>               [AC_MSG_RESULT(yes)],
>               [AC_MSG_RESULT(no)
>                LIBS="$cf_ncurses_SAVE"])
> fi
> 
> If the AC_TRY_LINK fails (as expected, for your configuration), the $LIBS
> variable is restored, with -lmytinfo.  (Seeing what that part of the log
> does might clear up the confusion).

Why you think that AC_TRY_LINK must fails? It NOT fails as expected, since
initscr() IS in ncurses() and it is dynamically linked thing. You should
call something like tigetstr() additionly to initscr() to make it really
fails without -lmytinfo. 

-- 
Andrey A. Chernov
<address@hidden>
http://www.nagual.pp.ru/~ache/

;
; To UNSUBSCRIBE:  Send a mail message to address@hidden
;                  with "unsubscribe lynx-dev" (without the
;                  quotation marks) on a line by itself.
;