Re: 23.0.50; WAS: strange messages on built-in variables

[Peter Dyballa]

Am 07.10.2007 um 02:50 schrieb Dan Nicolaescu:

> >         gcc -I/sw/include -L/sw/lib -DMAC_OSX -DHAVE_CONFIG_H
> > -I. -I../src -
> > I.../emacs/mac/../lib-src -I/Users/pete/
> > Quellen/Emacs_CVS/emacs/mac/../lib-src/../src -dead_strip -
> > bind_at_load -multiply_defined suppress -L/sw/lib/freetype219/lib -L/
> > sw/lib/fontconfig2/lib -L/sw/lib/ncurses -L/sw/lib -L/usr/local/lib -
> > g -O2 -Wno-pointer-sign   movemail.o pop.o getopt.o getopt1.o
> > -o movemail
> >         powerpc-apple-darwin8-gcc-4.0.1: getopt.o: No such file or directory
> > 	powerpc-apple-darwin8-gcc-4.0.1: getopt1.o: No such file or  
> > directory
> >         make[1]: *** [movemail] Error 1
> >         make: *** [lib-src] Error 2
>
> The Carbon port on CVS trunk has been abandoned by its maintainers and
> it is known to be broken. But the above is probably not related to
> what is known to be broken. If you are interested, can you try to
> figure out what goes wrong?

There seems to be something with src-lib/Makefile. When configuring  
and compiling the usual way, it goes:

        cd lib-src; make all \
          CC='gcc -...' CFLAGS='... ' CPPFLAGS='-DMAC_OSX ...' \
          LDFLAGS='...' MAKE='make'
        gcc ... -o test-distrib .../lib-src/test-distrib.c
        ./test-distrib .../emacs/lib-src/testfile
        gcc ... .../emacs/lib-src/make-docfile.c  -o make-docfile
        gcc ... .../emacs/lib-src/profile.c  -o profile
        gcc ... .../emacs/lib-src/digest-doc.c  -o digest-doc
        gcc ... .../emacs/lib-src/sorted-doc.c  -o sorted-doc
        cp .../emacs/lib-src/getopt_.h getopt.h-t
        mv getopt.h-t getopt.h
        gcc ... .../emacs/lib-src/movemail.c
        gcc ... .../emacs/lib-src/pop.c
        gcc ... .../emacs/lib-src/getopt.c
        gcc ... .../emacs/lib-src/getopt1.c
        gcc ...  movemail.o pop.o getopt.o getopt1.o         -o movemail

while the make-package script creates in mac a new build tree into  
which it cd's then invokes .../emacs/mac/../configure <options>. When  
configuration has finished, the script checks whether  .../emacs/ 
mac/../lisp/abbrev.elc exists – this is a flaw, because this file can  
exist from bootstrapping an X client, the Aqua-tic Emacs.app, or an  
MS Losedows application. If this ELC file exists, a 'make install' is  
tried. Even when instead of a 'make install' a simple 'make' is  
invoked, the compilation goes a bit different:

        cd lib-src; make all \
          CC='gcc ...' CFLAGS='... ' CPPFLAGS='-DMAC_OSX ...' \
          LDFLAGS='...' MAKE='make'
        gcc ... .../emacs/mac/../lib-src/make-docfile.c  -o make-docfile
        gcc ... .../emacs/mac/../lib-src/profile.c  -o profile
        gcc ... .../emacs/mac/../lib-src/movemail.c
        gcc ... .../emacs/mac/../lib-src/pop.c
        gcc ... movemail.o pop.o getopt.o getopt1.o         -o movemail
        powerpc-apple-darwin8-gcc-4.0.1: getopt.o: No such file or directory
        powerpc-apple-darwin8-gcc-4.0.1: getopt1.o: No such file or directory
        make[1]: *** [movemail] Error 1
        make: *** [lib-src] Error 2

test-distrib is not compiled and not executed, plus some others, and  
then the dependency that movemail depends on getopt.o and getopt1.o  
is not seen. The Makefiles in lib-src/Makefile are not that different:

        diff mac/make-package.build.10141/lib-src/Makefile lib-src/Makefile
        74,75c74,75
        < srcdir=.../emacs/mac/../lib-src
        < VPATH=.../emacs/mac/../lib-src
        ---
        > srcdir=.../emacs/lib-src
        > VPATH=.../emacs/lib-src
        78c78
        < top_srcdir=.../emacs/mac/..
        ---
        > top_srcdir=.../emacs


Right now I can't see a reason why GNU Make 3.80 behaves so  
differently ... because it sees that .../emacs/mac/../lib-src has the  
missing object files, although they're needed in .?

--
Greetings

  Pete

“One cannot live by television, video games, top ten CDs, and dumb  
movies alone”
      (Amiri Baraka 1999)