Bug#335392: Building OpenOffice for AMD64
Hi Rene,
Rene Engelhard wrote:
>
> rhino isn't in the .orig.tar.gz
That might explain.
> Anyway, I know how to fix this bug (have a patch for it) but fixing this
> bug for 2.0.0 is completely useless since it may build but not really work.
>
> That's why I don't put efforts into it.
ok.
> If you want a usable OOo on amd64 use the 32-bit one.
No, I want a true 64-bit version of OOo and I'm here to help.
> The ooo-build for 2.0.1 (and therefore the 2.0.1 packages) may build out of
> the box, though, but still not work.
Where can I find it ? Each time I do an "apt-get source openoffice.org",
I get the 2.0.0.
I found this page:
http://packages.debian.org/experimental/editors/openoffice.org
But how to specify in the "apt-get source" to grab this one ?
Is there plan, for Debian, to move quickly to the 2.0.1 ? A lot of
people are waiting for the AMD64 to be "official" and OOo is one major
software.
>>diff -ruN openoffice.org-2.0.0/ooo-build/patches/64bit/buildfix-64bit-config_office.diff openoffice.org-2.0.0-2/ooo-build/patches/64bit/buildfix-64bit-config_office.diff
>>--- openoffice.org-2.0.0/ooo-build/patches/64bit/buildfix-64bit-config_office.diff 2005-12-17 16:53:03.000000000 +0100
>>+++ openoffice.org-2.0.0-2/ooo-build/patches/64bit/buildfix-64bit-config_office.diff 2005-12-17 18:41:27.000000000 +0100
>>@@ -1,5 +1,14 @@
>> --- config_office/configure.in 2005-01-28 05:34:39.000000000 +0100
>> +++ config_office/configure.in 2005-01-28 05:36:21.000000000 +0100
>>+@@ -3357,7 +3385,7 @@
>>+ AC_MSG_RESULT([external])
>>+ SYSTEM_HUNSPELL=YES
>>+ AC_LANG(C++)
>>+- AC_CHECK_HEADER(hunspell.hxx, [],
>>++ AC_CHECK_HEADER(hunspell/hunspell.hxx, [],
>>+ [AC_MSG_ERROR(hunspell headers not found.)], [])
>>+ AC_CHECK_LIB(hunspell, main, [],
>>+ [ AC_MSG_ERROR(hunspell library not found.) ], [])
>> @@ -3481,17 +3481,17 @@
>> MOC="moc"
>> if test "$test_kde" = "yes" -a "$ENABLE_KDE" = "TRUE" ; then
>>diff -ruN openoffice.org-2.0.0/ooo-build/patches/64bit/intptr-basegfx.diff openoffice.org-2.0.0-2/ooo-build/patches/64bit/intptr-basegfx.diff
>
>
> Broken. Completely. Don't change a püatch shared by every distro with a
> debian-specific one. And anyway: debian-hunspell-build.diff already does
> the stuff necessary.
When I started to make the patches, this was'nt fixed and it was
aborting the compilation. I didn't know about the
debian-hunspell-build.diff patch, so this part of my patch is useless.
>>+++ openoffice.org-2.0.0-2/ooo-build/patches/OOO_2_0/apply 2005-12-17 18:44:04.000000000 +0100
>>@@ -32,7 +32,7 @@
>> DebianBase : LinuxCommon, SystemDB, DebianBaseOnly
>> Debian : DebianBase, DebianOnly, DebianSidOnly
>> DebianSarge : DebianBase, DebianOnly, DebianSargeOnly
>>-Debian64 : DebianBase, 64bit, DebianOnly
>>+Debian64 : DebianBase, 64bit, DebianOnly, IntPtr
>
>
> Which bars the question why IntPtr is not in the 64bit alias, but as
> that doesn't matter for 2.0.0 anyway...
The IntPtr patches should be applied on any architectures. It is about
removing errors of the type: I get a pointer out of an "malloc" and I
cast it to an sal_uInt32. Which is ok under 32-bits architecture but not
in 64-bits. Theses patches are safe.
> And you didn't explain why you disabled some patches...
Because they are not present in the archive and make the building
process fail.
Regards
--
Emmanuel Fleury
I'm just very selective about the reality I choose to accept.
-- Calvin & Hobbes (Bill Waterson)
Reply to: