I sent this question to fedora-list earlier, but it seemed to be the
wrong crowd or maybe the message just got lost in the noise.
Are there plans to provide OpenOffice.org 2.0 RPMs as updates to FC3
(I realize this is unlikely considering the status of GCJ in FC3)?
We are using OOo on a number of FC3 desktops, and I'd be interested in
upgrading to OOo 2.0 when it is released and still have OOo installed
locally as a RPM. We cannot do a wholesale upgrade to FC4 (more like
FC5 or later). /opt is on NFS on all machines, so vanilla 2.0 could
always go there, but then the users would likely have to learn about
one version in /opt and another in /usr (and the default FC Gnome
panel etc.). Updating the local RPM would be much better.
I tried installing the OOo packages from the development stream with
yum, which, of course, pulled in GCC 4 et al. with it. That version of
OOo seemed very unstable (Impress crashed almost immediately, whatever I
tried to do with it). I don't know whether that was because it is a beta
or because I was running FC4 development packages on FC3.
Are the release version packages from openoffice.org itself going to be
RPMs? The current beta ones are a bit odd, with the many 'core' RPMs and
files going in /opt. I guess the FC3 OOo RPMs could be replaced with
those, anyway, and the Sun JRE also be installed as a RPM.
Just noticed that the new gdm in rawhide allows early logon before the
entire system is up. But as I read about it in the init script does not
actually allow users to logon until the entire system is up. My question
is what is the purpose of this if you can't logon when the screen
We've been trying to pull together the components to make the release
notes an effort of the Fedora community. This means developers
collaborating with writers.
This will not mean what it has in the past, with developers throwing
some bits over the wall for release notes and sometimes reading and
commenting on the drafts. When Fedora had someone paid to write
relnotes, this worked. If we can't make this work as a community effort
with resources from writers _and_ developers, we may not have any
The first pass at an idea is on this page, scroll down to Release Notes
Process - Iteration One, April 2005:
The basic idea is to break the relnotes down into modules for individual
or teams of writers to tackle. Then you would have one or more writers
dedicated to a project or subset of Linux, such as kernel, network
services, printing, UI/desktop, SELinux, etc. Then it is easier to keep
the dedicated writing resource informed throughout the development
Hopefully this makes it possible for a volunteer writer to actually have
a chance to write an accurate relnotes piece in time for test or
I'd like to keep this part of the discussion f-devel-l, but have Cc:'d
f-docs-l so everyone is aware it. Reply-to is set to keep the
discussion on f-devel-list.
We're also debating the merits of single source in Wiki or DocBook/XML.
Advantages of the former are the ease of collaboration and getting new
writers able to contribute sooner. The advantages of XML are too
numerous to mention, but it does suffer from a longer learning curve.
If you have any comments on that, the tool choice discussion is
happening on f-docs-l. Regardless, we'll make sure the release notes
are in the proper format on the file system for Anaconda to use.
Karsten Wade, RHCE * Sr. Tech Writer * http://people.redhat.com/kwade/
gpg fingerprint: 2680 DBFD D968 3141 0115 5F1B D992 0E06 AD0E 0C41
Red Hat SELinux Guide
* Wed Apr 06 2005 Elliot Lee <sopwith(a)redhat.com> 10.2.0.45-1
- Deal with GUI-mode language traceback
* Wed Apr 06 2005 Elliot Lee <sopwith(a)redhat.com> - 10.2.0.44-1
- Deal with text-mode language traceback
- (clumens) Don't set SYSFONTACM
* Wed Apr 06 2005 Peter Jones <pjones(a)redhat.com> - 10.2.0.43-1
- Don't remove libraries in stage2 that don't match the one from linuxthreads/
* Wed Apr 06 2005 Jim Parsons <jparsons(a)redhat.com> 0.9.22-1.0
- Version Bump.
* Wed Apr 06 2005 Jim Parsons <jparsons(a)redhat.com> 0.9.21-1.0
- Changed path to lvm command for Fedora.
* Mon Feb 21 2005 Jim Parsons <jparsons(a)redhat.com> 0.9.20-1.0
- Added desktop icon file.
Just updated to FC4, test 1, latest packages to help out in bugspotting
I just noticed, that mysql isn't working in php 5. Is this known, and if
so, is it taken care of. If not, I would put some time into recompiling
* Tue Apr 05 2005 Jay Fenlason <fenlason(a)redhat.com> 2.4.4p4-4
- Add -bug144052 patch to close
bz#144052 amverifyrun sometimes verifies the wrong tapes
* Tue Apr 05 2005 Peter Jones <pjones(a)redhat.com> - 10.2.0.42-1
- Use linuxthreads libraries even if they're not the default, unless
explicitly told to use nptl
* Tue Apr 05 2005 Chris Lumens <clumens(a)redhat.com> 10.2.0.41-1
- Make sure $LANG is set right for the second stage.
- Fix kickstart traceback trying to skip a nonexistant step.
- Import encodings.idna (sopwith, #153754).
- Fix image building problems.
- Fix kickstart traceback when using shortened forms of language names
* Tue Apr 05 2005 Jason Vas Dias <jvdias(a)redhat.com> 3.1.8-70
- always call pam_close_session on PAM_FAIL or pam_end
* Tue Mar 08 2005 Jason Vas Dias <jvdias(a)redhat.com> 3.1.8-68
- Put PAM authentication check in 'check_permissions()', so
- user can know when using at(1) if PAM permission is denied.
* Tue Mar 08 2005 Jason Vas Dias <jvdias(a)redhat.com> 3.1.8-67
- better fix for bug 150131: change DAEMON_USERNAME and
- DAEMON_GROUPNAME to 'root' .
* Mon Apr 04 2005 Elliot Lee <sopwith(a)redhat.com> - 1.1.2-24
- Removed xmms-flac subpackage
* Wed Apr 06 2005 Jakub Jelinek <jakub(a)redhat.com> 4.0.0-0.40
- update from CVS
- PRs ada/19956, c++/18644, c++/19159, c++/19199, c++/20679, debug/20505,
libfortran/20068, libfortran/20125, libfortran/20156,
libfortran/20471, libfortran/20755, libgcj/20727, libgcj/20761,
rtl-optimization/16104, rtl-optimization/20527, target/19537,
- fixed ABI bug where __attribute was ignored on certain struct
fields (Mark Mitchell, PR c++/20763)
- don't error on #define a! in c89 pedantic mode (PR preprocessor/19475)
- prevent inlining functions that call __builtin_apply_args or
__builtin_return (PR tree-optimization/20076)
- add g++.dg/compat/struct-layout-1.exp compatibility tests
* Mon Apr 04 2005 Jakub Jelinek <jakub(a)redhat.com> 2.3.4-21
- update from CVS
- fix xdr_rmtcall_args on 64-bit arches (#151686)
- fix <pthread.h> and <bits/libc-lock.h> with -std=c89 -fexceptions (#153774)
* Tue Apr 05 2005 Ray Strode <rstrode(a)redhat.com> 1:2.10.0-5
- Don't use %postun -p optimization now that we do more than
just /sbin/ldconfig in %postun (bug 152236)
* Tue Apr 05 2005 Than Ngo <than(a)redhat.com> 3.4.0-2
- xmms is removed in fc4, rebuild without xmms support
* Tue Apr 05 2005 Than Ngo <than(a)redhat.com> 7:3.4.0-3
- xmms is removed in fc4, rebuild without xmms support
* Tue Apr 05 2005 Peter Jones <pjones(a)redhat.com> - 4.2.8-1
- Don't consolidate duplicates in "--args" (#147222)
* Tue Apr 05 2005 Radek Vokal <rvokal(a)redhat.com> 1.60-51
- flush output in mii-tool (#152568)
* Mon Apr 04 2005 Elliot Lee <sopwith(a)redhat.com> - 1:1.9.89-3
- Slim down the langpacks
* Sat Apr 02 2005 Caolan McNamara <caolanm(a)redhat.com> 1:1.9.89-2
- package the wizard jars (doh!)
- fix the fpic icudata issue properly
- enthused by parallized build success try parallized build.pl as well as dmake
- some problems during langpack creation -> maxprocess issues ?
- damn ppc gives illegal instruction on sysui mandrake menus which it didn't
in the last build. I couldn't be bothered wasting my time figuring out why
* Tue Apr 05 2005 Will Cohen <wcohen(a)redhat.com>
- Backport ppc64 patch for synthesizing dotted symbols.
* Tue Apr 05 2005 Dan Walsh <dwalsh(a)redhat.com> 2.0.0-2
- Fix buildrequires lines in spec file
* Wed Mar 02 2005 Dan Walsh <dwalsh(a)redhat.com> 2.0.0-1
- Update to latest from tresys
* Mon Nov 29 2004 Dan Walsh <dwalsh(a)redhat.com> 1.5.1-6
- add FALLBACK=true to /etc/security/console.apps/apol
* Tue Apr 05 2005 Paul Nasrat <pnasrat(a)redhat.com> - 1.2.11-1
- Fix traceback
* Wed Mar 09 2005 Jens Petersen <petersen(a)redhat.com> - 8.3.5-6
- add unversioned symlinks to the static libs (Dave Botsch, 149734)
- rebuild with gcc 4
- add tclx-8.3.5-clock_t-gcc4.patch to skip clock_t test in configure
- buildrequire autoconf213
* Tue Apr 05 2005 Jason Vas Dias <jvdias(a)redhat.com> - 4.1-28_FC4
- Required for EAL Audit certification:
- If pam_setcred should fail, the pam_session could fail to be
- closed, leaving autofs user directories still mounted.
* Mon Apr 04 2005 Mike A. Harris <mharris(a)redhat.com> 4.0.17-1
- Updated to new xdm-config from xorg-x11-6.8.2
- Change xdm-config to invoke Xwilling with "-s /bin/bash" instead of "-c"
to fix bug (#86505)
* Mon Apr 04 2005 Mike A. Harris <mharris(a)redhat.com> 4.0.16-2
- Added mkdir -p $RPM_BUILD_ROOT/etc/X11/xinit/Xclients.d to install section
* Mon Apr 04 2005 Mike A. Harris <mharris(a)redhat.com> 4.0.16-1
- Added /etc/X11/xinit/Xclients.d directory and enhanced Xsession to use it,
as a solution for (#112478,122941,142260)
- Updated Xclients with fixes for fallbacks and various general cleanups
for bug (#139285)
- Removed hard dependancy on switchdesk, as it is no longer required
* Mon Apr 04 2005 Mike A. Harris <mharris(a)redhat.com> 6.8.2-19
- xfs.init updated with fixes for chkfontpath invocation, as the previous
code did not filter the output of chkfontpath very good. The new sed
filtering is more reliable now, until chkfontpath is enhanced to provide
the info in a way that does not require filtering.
- xfs.init updated to handle .ot[cf] fonts before .tt[cf] fonts, which is a
small optimization for the case of dirs containing .ot[cf] and .tt[cf]
fonts, as it bypasses the prior invocation of ttmkfdir which would get
discarded anyway. Now only one of the two will be invoked, however the
resulting fonts.dir will be identical to before.
- xfs.init updated to *always* run fc-cache without arguments, in order to
ensure all fontconfig configured font directories have updated cache
- xfs.init should no longer try to create files on read-only filesystems,
coincidentally due to the above changes, except when the font metadata
is genuinely outdated, which is an error condition no matter how the
filesystem is mounted (#74398)
I am the RULE project coordinator. I have never formally introduced
myself here because I don't plan to become a regular FC developer, I
just lurk on this list to know what will happen in Fedora. This time,
however, I think I can provide some useful information:
On Wed, Apr 06, 2005 21:32:00 PM +0100, Andy Green (andy(a)warmcat.com)
> Alan Cox wrote:
> | There have been tools like miniconda.
Miniconda was the first RULE installer. It is a patched anaconda and
is NOT maintained anymore by us. The current one is slinky, see
below. Feel of course free to port miniconda to FC4 if you think it
> | I've been down this path a bit and you hit a single fundamental
> | near-insoluble problem - up2date/yum/rpm. They burn resources and
> | need some major rework at the rpm level
One generic comment on this: even if someone rewrote from scratch the
whole up2date/yum or apt4rpm/ rpm chain, I do NOT think it would help
on limited systems: it would just tell you much faster that you can't
install because you have no disk space.
The package management system can do little when the code of an
application was originally written (or sometimes packaged later) to
require any other package in sight to just use one or two files from
it, and that other package has the same limit.
When we started RULE I looked into some ways to automatically spot out
the worst/heavier packages and package sets from this point of view,
but frankly had not enough skills. I would *really* like to cooperate
with some real developer on this. Basically, apt, yum and friends just
tell you "don't worry, I will fill the HD with possibly unnecessary
packages without wasting your time". I would like to see/help with the
development of something like the DAN I tried to do back then:
> Busybox has a really really minimal core re-implementation of
Yes, this is how slinky, the current RULE installer works. It uses
busybox and its rpm to install the real rpm, and then the latter to
pull in manually crafted package lists. Originally, DAN was supposed
to generate these lists automatically at each new distro release.
Again, I think something like DAN would help a LOT also with packaging
and maintenance of standard FC, so it would be great to work on it.
Anyway, the current RULE installer is shortly explained here:
FC3 versions are here
and their description is in this announcement:
Slinky is the only one we are working on. Help on it and/on DAN is
Marco Fioretti mfioretti, at the server mclink.it
Fedora Core 3 for low memory http://www.rule-project.org/
Do not ascribe to malice that which can be adequately explained by
stupidity. Physicist Richard Feynmann, Nobel Prize
I'm curious to know what types of configuration developments are
happening around the MergedFB mechanism in X. Currently, I can share a
dual head configuration and a single head configuration together except
for two options:
1a. Option "MetaModes" "1024x768-1280x1024 1024x768"
b. Option "MetaModes" "1024x768"
2a. Virtual 2304 1024
b. ---not defined---
a is for my dual head setup and b is for my single head setup.
Obviously, my target of concern is a laptop with an internal panel as
one head and an external monitor as the other.
Right now, I have to change these manually. Ctrl_Alt_+ and Ctrl_Alt_-
are a *pain* to compose on my IBM T30. And, frankly, with the virtual
at 2304 1024, I'm not certain how much that'd help me anyway.
So, is there an RandR interface into MergedFB and would
system-config-display have control over it? Does system-config-display
support MergedFB setups when configuring dual head? Is this on the
(I know all cards do not support it, but it would be nice if a
capabilities mechanism were able to detect this based on the card's
vendor string, etc.)
Thanks for any links, hints, and your ideas on this...
How close will FC4's Perl, be to the Perl 5.8.6 release?
T +44 (0) 1224 279484
M +44 (0) 7930 323266
F +44 (0) 1224 742001
Open Source. Open Solutions(tm).
I installed FC4 test1 and xen0 straight from Fedora and ran some simple
kernel microbenchmarks (loops over getpid, fork/wait, signal, etc.) and the
results are pretty ugly compared to a 2.6.1 install. FC4 ranges from 1.5X to
20X slower, while xen0 is about 2X to 6X slower. Is this expected? I'm
looking for configuration differences now, but I'm wondering if anybody can
comment on how FC4 and xen0 were built. debug? asserts? One difference
between FC4 and xen0 is that FC4 was configured with HIGHMEM and xen0 wasn't.