[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Building the packages for Woody (1.0.1-6)



Jerome Warnier wrote:
Chris Halls wrote:

On Mon, Nov 11, 2002 at 01:56:26PM +0100, Jerome Warnier wrote:
Any plans to backport to Woody?


Of course, but the word 'unstable' is carefully chosen.. I will only think
about doing a backport once the dust settles, and that will probably be
after I am reasonably happy that no new problems have arisen and have made
an upload to unstable.

This version brings a lot of bug corrections that affect also Woody.
I now have machines with enough disk space and power to compile it myself, but this version is not "buildable" as such.

I'm trying to do it myself, but I run into some (small) problems:
- why is a newer debhelper version needed? (is it really needed?)
- do we really have to backport g++-3.2 to Woody, or is there another solution?


Actually, it is more "buildable" than the previous version, because the
stlport problems have gone. It is best to build with gcc 3.0 so that you do
not need to backport 3.2.  Here is a quick guide:

- Install the debhelper from here:
 http://ftp.freenet.de/pub/ftp.vpn-junkies.de/openoffice/testing/
 It is needed because the version of dh_install in Woody has a bug.  This
 package backports the 2 line change that is needed to fix the problem.

- Put this patch into debian/patches:

http://cvs.debian.org/*checkout*/oo-deb/debian/patches/Attic/006_exception_sprecs.diff?rev=1.3&cvsroot=debian-openoffice&hideattic=0

- Edit debian/control and change 'g++-3.2' to 'g++-3.0'

- Edit debian/scripts/vars.i386 and change 'GCC3VER=3.2' to 'GCC3VER=3.0'

- Build :)

Done all that.


Done all this, but I still end up with different problems:

The first one is a "permission denied" on excvpe for java_detect.sh.
How is this, I still have this problem. It is just this script not being executable, I fixed it by hand (chmod +x) after interrupting the build process.



The second one is that it asks for a strange question.
Here you have the log:

********************************************************************
*                                                                  *
*   Setting up the build environment variables.                    *
*                                                                  *
********************************************************************

$COMPATH=NO_COMPATH
$_gcc_include_path=/usr/lib/gcc-lib/i386-linux/3.0.4/include
$_gxx_include_path=NO_GXX_INCLUDE
$JAVA_HOME=/usr/lib/j2se/1.3
$TCSH=/usr/bin
$PERL=/usr/bin
$x_libraries=/usr/X11R6/lib
$x_includes=/usr/X11R6/include
$_LOCAL_SOLENV=DEFAULT
$_LOCAL_SOLVER=DEFAULT
$upd=641
$CYGWIN=NO_CYGWIN
$STLPORT4=NO_STLPORT4
$enable_xprint=TRUE
$with_lang=,ALL,
$with_asm_home=NO_ASM_HOME
$with_unzip_home=NO_UNZIP_HOME
$MINGWIN32=0
$with_use_shell=tcsh
$USE_GCC3=TRUE
$CC=gcc-3.0
$CXX=g++-3.0

Setting up the environment for building OpenOffice.
Setting platform independent values... done
Setting Linux Intel specific values... done
NO_GXX_INCLUDE = no_gxx_include, NO_GXX_INCLUDE = argv(2)
Please enter the g++ include path (e.g. /usr/include/g++-3):

On my system, I happen to have a /usr/include/g++-3 and a /usr/include/g++-v3 directories.
Which one should I choose?
Why isn't that automated?

I still have this problem too. Where does it come from?
Is it Woody-related?

This time, I tried with gcc-3.0 and ccache.

Is it possible to build in parallel now? Is there still problems with the build process? It would go *really* faster in my case, because I could use a lot of machines in // (maybe 20!).

So far it did not finish, but those bugs are still there.
I'll let you know tomorrow.

There may be new problems - I haven't tested them yet - but that should
solve most build problems.
Here you have some more to keep you busy ;-)

Chris





Reply to: