Re: compiling bison-3.0 on AIX 5.3

[Michael Felt]

repeating the process, after removing all the old, worked as far as make
dist is concerned.

The new dist compiled correctly on AIX 5.3, but make check failed due to no
LEX I assume.
address@hidden:[/data/prj/gnu/bison/bison-3.0.8.1]buildaix
+ CPPFLAGS=-I/opt/include CFLAGS=-O2 -qlanglvl=extc99 ./configure \
        --prefix=/opt \
        --sysconfdir=/var/bison/etc \
        --sharedstatedir=/var/bison/com \
        --localstatedir=/var/bison \
        --mandir=/usr/share/man  \
                > build/aix/configure.out
cmp: EOF on conftest.ok
+ make > build/aix/make.out
extexi:  GEN examples/rpcalc/rpcalc.y
extexi:  GEN examples/mfcalc/mfcalc.y
extexi:  GEN examples/mfcalc/calc.h
extexi:  GEN examples/calc++/calc++-driver.hh
extexi:  GEN examples/calc++/calc++-driver.cc
extexi:  GEN examples/calc++/calc++-parser.yy
extexi:  GEN examples/calc++/calc++-scanner.ll
extexi:  GEN examples/calc++/calc++.cc
    1500-030: (I) INFORMATION: gram_lex: Additional optimization may be
attained by recompiling and specifying MAXMEM option with a value greater
than 8192.
+ make install DESTDIR=/var/tmp/root/bison-3.0.8.1 > build/aix/install.out
+ mkinstallp.ksh /var/tmp/root/bison-3.0.8.1 > build/aix/mkinstallp.out
==============================
aixtools.bison:aixtools.bison.man:3.0.8.1::I:T:::::N:bison man pages
2013.Aug.25 11.49 CUT::::0::
aixtools.bison:aixtools.bison.rte:3.0.8.1::I:T:::::N:bison 3.0.8.1
2013.Aug.25 11.49 CUT::::0::
aixtools.bison:aixtools.bison.share:3.0.8.1::I:T:::::N:bison share and
documentation 2013.Aug.25 11.49 CUT::::0::
==============================
address@hidden:[/data/prj/gnu/bison/bison-3.0.8.1]make check
        make  check-recursive
Making check in po
Target "check" is up to date.
Making check in runtime-po
Target "check" is up to date.
Making check in .
        make  examples/calc++/calc++ examples/mfcalc/mfcalc
examples/rpcalc/rpcalc ./tests/bison tests/atconfig tests/atlocal
  YACC   examples/calc++/calc++-parser.stamp
  CXX    examples/calc++/examples_calc___calc__-calc++-driver.o
  LEX    examples/calc++/calc++-scanner.cc
make: 1254-004 The error code from the last command is 1.


Stop.
make: 1254-004 The error code from the last command is 2.


Stop.
make: 1254-004 The error code from the last command is 1.


Stop.
make: 1254-004 The error code from the last command is 2.


Stop.
address@hidden:[/data/prj/gnu/bison/bison-3.0.8.1]



On Sun, Aug 25, 2013 at 1:24 PM, Michael Felt <address@hidden> wrote:

> Thanks for the suggestion to use git, etc, on Linux.
>
> needed to install packages: in addition to what you listed above to
> verify/install: autopoint, "makeinfo" via texinfo and help2man.
>
> All looked fairly well, the initial make complained about help2man not
> being available (at the end), and consequently, make dist did not complete.
>
> After running 'make maintainer-clean' - as suggested, after ./configure
> (even after a new bootstrap and configure) make fails to run to end
> stopping with message:
>   YACC   src/parse-gram.c
> /data/prj/gnu/bison/bison/./tests/bison: line 28:
> /data/prj/gnu/bison/bison/src/bison: No such file or directory
> Can't open y.tab.[ch]: No such file or directory.
> make: *** [src/parse-gram.c] Error 1
>
> === make V=1 output ===
> address@hidden:/data/prj/gnu/bison/bison# make V=1
> /bin/bash ./build-aux/ylwrap src/parse-gram.y y.tab.c src/parse-gram.c
> y.tab.h src/parse-gram.h y.output src/parse-gram.output -- ./tests/bison -o
> y.tab.c -d -v -Werror -Wall -Wno-yacc --report=all
> /data/prj/gnu/bison/bison/./tests/bison: line 28:
> /data/prj/gnu/bison/bison/src/bison: No such file or directory
> Can't open y.tab.[ch]: No such file or directory.
> make: *** [src/parse-gram.c] Error 1
> =======
> address@hidden:/data/prj/gnu/bison/bison# ls src
> AnnotationList.c  getargs.h         main.c           print-xml.h
> scan-skel.h
> AnnotationList.h  gram.c            muscle-tab.c     reader.c
> scan-skel.l
> assoc.c           gram.h            muscle-tab.h     reader.h       state.c
> assoc.h           graphviz.c        named-ref.c      reduce.c       state.h
> closure.c         graphviz.h        named-ref.h      reduce.h
> symlist.c
> closure.h         ielr.c            nullable.c       relation.c
> symlist.h
> complain.c        ielr.h            nullable.h       relation.h
> symtab.c
> complain.h        InadequacyList.c  output.c         Sbitset.c
> symtab.h
> conflicts.c       InadequacyList.h  output.h         Sbitset.h
> system.h
> conflicts.h       lalr.c            parse-gram.[ch]  scan-code-c.c
> tables.c
> derives.c         lalr.h            parse-gram.y     scan-code.h
> tables.h
> derives.h         local.mk          print.c          scan-code.l
> uniqstr.c
> files.c           location.c        print_graph.c    scan-gram-c.c
> uniqstr.h
> files.h           location.h        print_graph.h    scan-gram.h
> flex-scanner.h    LR0.c             print.h          scan-gram.l
> getargs.c         LR0.h             print-xml.c      scan-skel-c.c
>
> =======
> The first time make seemed to go fine. Will delete the whole git
> directory, and restart - so this is FYI
>
>
> On Sun, Aug 25, 2013 at 3:49 AM, Paul Eggert <address@hidden> wrote:
>
>> Michael Felt wrote:
>> > do I need to do something besides ./configure
>> > (I did "make distclean", then configure, then make V=1)
>>
>> If you're building from git, please follow the instructions in
>> README-hacking.  Here's a precis:
>>
>> * Make sure you have up-to-date versions of autoconf, automake, flex,
>>   gettext, git, graphviz, gzip, perl, rsync, and tar installed.  If this
>>   isn't feasible, do the remaininng steps on a GNU/Linux host that
>>   has recent tools installed, then type 'make dist', and copy
>>   the distribution tarball to AIX and build the tarball as usual.
>>
>> * Run the following shell commands:
>>
>> git clone git://git.savannah.gnu.org/bison.git
>> git checkout maint
>> git submodule update --init
>> ./bootstrap
>> ./configure
>> make
>> make check
>>
>> It'll take a while; sorry.  If you're on a multicore machine
>> something like 'make -j8' should be faster than plain 'make'.
>>
>
>