DevHeads.net

Currently I am following F13 and I have been noticing a lot of packages
disappearing from updates-testing before showing up in the branched release
(but similar things happen with normal updates, just less often).

When things disappear from updates-testing it isn't immediately obvious
if this is because the updates are bad or because the update has been pushed
to stable.

Since I have a local mirror and the packages disappear between syncs, I end
up downloading them twice.

I got a message about cegui having a broken dependency in devel that didn't
seem to make any sense since all of the dependencies in the package referred
to subparts of the package having matching versions. I hadn't changed it
for about a week, so it seemed suspicious that it showed up the day
after the branch.

I saw that there is a change in the way git push works in 1.7.
Currently I only do simple things and a typical workflow is:
git pull
Make a few changes
git commit -a
git push

This normally does a fast forward update and avoids a merge.
Is there another way I should be doing this? Does the answer change for 1.7?

It is hard to do testing for spins when there is also some broken dependency
in rawhide. (The latest last minute breakage being gmime.) Doing soname
bumps right before the alpha freeze (I think under the new development rules
the updates would be blocked for beta and beyond) results in fire drills
with people scrambling to get packages rebuilt and makes spin testing
difficult.

Perhaps a separate freeze date for soname bumps (at least for those that are
dependencies for other packages) could be made.

Ben did some updates tonight for rawhide and I am able to play tremulous now
on an nVidia nv28 based card.

It's pretty late so I didn't do much testing and there may be bugs. But if
you have nv20 series or earlier cards you might want to give it a try
with Monday morning's rawhide.

The following article has information about Nouveau getting working 3d
support for older cards (up to some nv20s) very soon:
http://www.phoronix.com/scan.php?page=news_item&px=Nzk1NA

Since I have some scrounged machines that came with nv20s, I am very
pleased by this. This will make testing the games spin easier.

I am working on getting squashfs-tools 4.1 in rawhide. It has wrapper
functions that are set up to use streaming compression/uncompression
from the LZMA SDK 4.65 library.

Currently the LZMA SDK 4.32 library is in Fedora, but is no longer supported
upstream and is not compatible with 4.65.

I think the xz-devel and xz-libs packages support a 4.65 compatible format
using the LZMA1 filter, but I am having trouble figuring out how to use
this library.

I don't think packaging the LZMA SDK 4.65 library is a good option.

I just went to rawhide over the last day and am not able to boot into
kernel 2.6.32-0.48.rc7.git1.fc13 unless selinux is disabled. (permissive
isn't good enough). I can boot into my old kernel 2.6.31.5-127.fc12 which
had a dracut generated image from before the upgrade. The error occurs
when udev is trying to unlock my nonroot partitions. I get an error
message refering to filesetcon not working on a /dev/mapper file. I get
asked for passwords again (since all of the file systems have the same
luks password I normally don't have to do this) and the correct password
doesn't work.

We currently have a 3d chess game packaged as chess. I want to ask for
fedora hosted space for it sop that we can be upstream for some modernization
(with regard to ogre and gcc) changes. However it strikes me that the package
name probably shouldn't have been 'chess' as that is a generic name. Before
officially asking for fedora hosted space I wanted to see if there is
sentiment to making some name change for F13+ and fedora hosted. Possibilities
would be 3dchess (possibly still too generic), ogrechess (possibly misleading,
as people might think it had ogres for pieces), chess-ogre or ?

I am working on packaging pagedgeometry and I noticed that when building
on gcc it passes -msse which I am guessing says to use sse instructions.
I think that even in F12 we can't assume these instructions are available.
The package may gain a lot of benefit from using those instructions.
(I haven't tested that yet as I am still pretty early in the process.)
Is there some relatively standard way to handle something like this?

blobAndConquer seems to have dropped out of today's rawhide repo even though
it is still tagged for dist-f12, f12-beta and f12-final.
Are other people seeing this?

I am wondering if there is any documentation on tips for packagers for
helping them integrate sound features of their packages in Fedora.

In my specific case the volume controls on glest are not affecting the
volume in F12 and I am wondering where to start looking for infomation
that may be relevant for fixing this.

On Fri, Sep 25, 2009 at 07:52:35 -0700,

I looked through the man page and tried a couple of things that seemed might
work and found the -P 2 option got this query to work.

Quite a while ago I asked the list for headset recommendations for chatting
and listening to music by someone who doesn't need great sound. This is a
followup to thank people for their suggestions and to note what I ended up
getting.

Several people recommended Logitech's clearchat commfort USB headsets.
That's what I went with (2 for home, 1 for work). I tried them on F11
and F12 and the sound, mic and inline volume control all work.

I found the following quote from a LWN article insteresting:
Internally, kernel.org runs on ten "disgustingly nice" machines donated by HP. John was strong in his praise of HP and ISC (which provides the bulk of the considerable bandwidth used by kernel.org); without them, kernel.org would not function the way it does. Beyond ISC, there are a couple of machines hosted at the OSU open source lab and one at Umeå University in Sweden. A lengthy process has finally gotten all of these machines upgraded to Fedora 9 - just in time, John noted wryly, for Fedora to end support for that distribution.

epiphany-2.26.3-2.fc11 in koji has been built against the newer gecko-libs
but wasn't pushed to bodhi (in either updates or updates-testing). People
who have it installed are not going to get the firefox related updates
that were recently pushed to updates without manual intervention.

Will I just be able to use the rawhide versions of Postgres 8.4 rpms (when
they show up) in F11 or will I need to rebuild from the source package?

I am the maintainer of the Live Games Spin and am looking for feedback
on dealing with games (though this could apply to other types of packages
as well) that have some of their content obtained using autodownloader.
Currently I don't want to include games that can only be played using
downloaded content on the Live Games Spin. I don't think that gives
a good experience using the Live Games Spin as a demo (which is what
I think its focus should be).

Our Spins SIG meetings have been pretty poorly attended recently (just nirik
and I) and our Spins Wrangler for F11 has resigned from that position for F12.

nirik does not have time to take a more active role and I want to limit
my work in the Spins SIG to technical stuff and the Games Spin.

I spent some time this past weekend working on getting fedora talk working
with asterisk.

Twice now I have seen maintainers confused about how to specify release
strings in an obsoletes statement. They like to drop the .fc11 and then
use the version and the rest of the release string from the exact version
they want to obsolete. And this won't work.
Is there some place to more prominently document this?
Is this something rpmlint catches or could be made to catch?

In order to work around the current dbus issue I copied my repo mirror
to a spare hard drive on the machine. I strip out some of the stuff in
the image directory to save space, but I found that the install wants to
use the install image file from the repo on the hard drive rather than
the one on the boot disk (boot.iso).
Is there a good reason for this?

I got an error message during the install of gnubg while running livecd-creator
to build a games spin.

While trying to make a custom games livedvd I saw some script errors
and was wondering if those were packaging errors or if livecd-creator
is doing something wrong.

I have a really old bug (163701) for a problem with my DVD/CD burner that
once upon a time used work with Fedora. Then for a while it would only
work after first erasing a CD until the next reboot. Now it no longer
burns CDs at all (but burning DVDs works).
The bug languishes (possiblly because it is hardware dependent and people
don't have that hardware and that isn't likely to change for the better
going forward) and one thing I thought of that might account for the
change in behavior is that wodim issues some commands that are being filtered
by the kernel.

I had an issue with gcin running when it input methods were supposed
to be turned off (if I can duplicate that part of the problem I'll file
a bug report on it) and it was causing a noticeable lag between when I would
hit a key and when the character would show up. Is it normal for it to have
a delay on the order of a tenth of a second when using gcin? If not I'll
retest it and also make a report about that. I suspect the behavior is
necessary in order for key combinations to work, but am not sure.

I am seeing some internal compiler errors when building svn versions of
Wesnoth this evening using gcc-4.3.2-5.i386. I wasn't seeing that last
night.
I have added the output recommended by gcc to bug 466198.
I don't know how fast new compiler versions start getting used to build
other stuff, but it case its right away I wanted to put out a heads up.

I am trying to use repoquery to return a list of packages that provide any
of a list of strings. For example something like:
repoquery --disablerepo='*' --repofrompath=i386,/home/rawhide/koji/i386/ --enablerepo=i386 --whatprovides `cat temp` > temp2

The above sort of seems to work but it looks like it is searching either the
installed packages or one of the other repositories. I tried using
--pkgnarrow, but that didn't seem to help.

I am playing with making custom repos combining the latest rawhide with the
latest koji updates.

Something changed /etc/sysconfig/iptables back to what looks to be the
default initial rules within the last couple of days. rpm shows that file
as unowned (no packages provide it). iptables doesn't appear to have been
updated recently so i don't think an update did that. Is the firewall state
saved someplace else?
I did use the gui selinux configuration interface, but not (knowingly) the
gui firewall interface. Normally I use service iptables save command when
I want to save something complicated or just edit the file directly when
it is a simple change.