(no subject)
by Ricardo Ariel Gorosito
On 5-July Thomas Fitzsimmons wrote:
"The java command will be a wrapper script that runs gij. When gij is
invoked on a class name, it first searches in /usr/lib for a
natively-compiled version of that class."
Is this jdkgcj? when be incorporated in fedora?
Ricardo.-
19 years, 10 months
/sbin/install-info causing scripts to fail esp. in gcc srpms
by Robert Hardy
Hello, People
I'm a System Engineer who has been quietly (re)building distributions for
many years now (since early 1990s.)
There seems to be a rare but reoccuring bug w/ install-info that seems to
occur a hugely disproportionate amount in the post/pre scripts of gcc srpms.
The earliest instance I recall was with libgcj-3.2.1-2 back in Redhat 8.0 and it is
present again in the latest devel gcc-gnat-3.4.1-7.
In gcc-gnat's case, install-info seems unable to delete gnat_ugn_unw from
/usr/share/info/dir as we see here:
/sbin/install-info --delete --info-dir=/usr/share/info /usr/share/info/gnat_ugn_unw.info.gz
install-info: warning: no entries found for /usr/share/info/gnat_ugn_unw.info.gz'; nothing deleted
FYI shell returns 0
However it does realize it is in fact there if I try to reinstall it:
/sbin/install-info --info-dir=/usr/share/info /usr/share/info/gnat_ugn_unw.info.gz
install-info: menu item `GNAT User's Guide (gnat_ugn_unw) for Native Platforms / Unix and Windows
' already exists, for file (none)'
FYI shell returns 1
The grep below confirms that the file is really still listed in the dir file.
grep -i gnat_ugn_unw /usr/share/info/dir
* GNAT User's Guide (gnat_ugn_unw) for Native Platforms / Unix and Windows
If I change gcc-gnat spec file's post gnat script to somelike like this it
fixes the dir file and avoids the problem:
%post gnat
/sbin/install-info \
--info-dir=%{_infodir} %{_infodir}/gnat_rm.info.gz
if ! grep -q '(gnat_ugn_unw)' /usr/share/info/dir; then
grep -v '(gnat_ugn_unw)' /usr/share/info/dir | >/tmp/infodir.$$
mv /tmp/infodir.$$ /usr/share/info/dir
fi
/sbin/install-info \
--info-dir=%{_infodir} %{_infodir}/gnat_ugn_unw.info.gz || true
Any time one of these messes occurs we end up with two different version of
a given rpm in the database and are forced to remove both versions, remove
the entry which didn't get deleted properly from /usr/share/info/dir and
re-install the package.
I'm not trying to say libgcj or gcc-gnat are not important, however if this
happened with a more critical system library such as libstdc++ or glibc this
could get exceedingly ugly.
In my opinion, we really should not be failing to upgrade/install critical
tools because their info files fail to get indexed properly.
Anyone know why this keeps happening?
Could we PLEASE do something to make the failure of a /sbin/install-info
install only a warning (it is very painful to have to keep recompiling gcc
to fix this issue)?
Adding || true to the end of ALL install-info install commands (--delete
lines don't need it) in the gcc rpm seems to guarantee we will avoid this
problem. For example:
/sbin/install-info --info-dir=/usr/share/info /usr/share/info/gnat_ugn_unw.info.gz || true
Could someone please do this in the devel gcc spec for now at least until
the real problem is found?
Regards,
Rob
--
---------------------"Happiness is understanding."----------------------
Robert Hardy, B.Eng Computer Systems C.E.O. Webcon Inc.
rhardy <at> webcon <dot> ca GPG Key available (613) 276-7327
19 years, 10 months
Modem SM56 Motorola
by Hugo Muñoz
Hi, I have a Motorola SM56 PCI modem and y want to know if there is an available driver for this modem on fedora, I find out one for Red-Hat 9 but I can't find one for Fedora I need this Urgent please help me.
PD: Sorry about my English.
19 years, 10 months
s390 ctc0 interface not coming up
by Matthias Saou
Hi,
Using hercules, I managed to get Fedora Core Development for s390 installed
and running. My remaining issue is that the ctc0 network interface doesn't
come up in the installed system, although it got configured fine and got
used at install time. The network service reports "OK", but the interface
doesn't appear.
Here's the content of my ifcfg-ctc0 file, it's the post-install default and
seems correct to me :
# IBM CTC
DEVICE=ctc0
BOOTPROTO=static
BROADCAST=192.168.200.3
IPADDR=192.168.200.3
MTU=1500
NETMASK=255.255.255.255
NETWORK=192.168.200.3
ONBOOT=yes
QETH=yes
REMIP=192.168.200.4
SUBCHANNELS=0.0.0600,0.0.0601
TYPE=QETH
And the last lines of the dmesg output (nothing regarding ctc0...) :
[...]
ip_tables: (C) 2000-2002 Netfilter core team
NET: Registered protocol family 10
Disabled Privacy Extensions on device 002155f4(lo)
IPv6 over IPv4 tunneling driver
divert: not allocating divert_blk for non-ethernet device sit0
Is this a known problem? I could investigate some more, but it's just so
annoying to have to prefix all commands with dots and not having command
completion... if only I got the network working! ;-)
The last thing I tried was modprobing the ctc module (I checked, it has no
options itself, dunno where the SUBCHANNELS stuff is needed) and doing a
"service network restart", but no better even though it reports "OK" for
ctc0, and ifconfig on ctc0 still reports that the device isn't found.
Matthias
--
Clean custom Red Hat Linux rpm packages : http://freshrpms.net/
Fedora Core release 2 (Tettnang) - Linux kernel 2.6.6-1.435.2.3
Load : 0.09 0.32 0.33
19 years, 10 months
fedora.us downtime this weekend
by Warren Togami
Friday July 23, 2004 around 4:00pm HST the fedora.us buildsystem and
bugzilla will be offline until late Saturday or early Sunday HST. Power
will be out in the University of Hawaii building that hosts the servers,
so we have no choice here. Temporarily download.fedora.us will be
repointed to Pix's server in France, and I am currently working on
moving the main fedora.us webpage to a temporary server elsewhere.
During this downtime the repository will remain unchanged until power
comes back. Mirrors syncing against the main server in Hawaii should
probably disable their rysnc in cron until services are restored.
Warren Togami
wtogami(a)redhat.com
19 years, 10 months
Importing third-party developer's public keys
by Michel Salim
Hello,
When QA-ing packages from non-core developers (i.e. those whose public
keys are not shipped with fedora-rpmdevtools) is there a way to add
the keys manually?
I checked fedora-installdevkeys and it seems to just perform rpm
--root ~/.fedorarpm --import PUBKEY; so after downloading the
developer's public key (gpg --recv-keys 1b4259b3 ; gpg --export
--armor 1b4259b3 > PUBKEY) I did just this.
rpmlint'ing or fedora-rpmchecksig'ing the .src.rpm kept giving a
MISSING KEY warning though. What did I do wrong?
Thanks,
- Michel
19 years, 10 months
Yet another FC3 wishlist
by Michel Salim
Hi,
Would it be possible to include support for the Intel Pro Wireless
2100 and 2200 wireless cards? (The drivers are available at
ipw2100.sf.net and ipw2200.sf.net respectively)
Most Centrino laptops come with either of these cards, so it would be
a nice addition. There might be license problems with bundling the
binary firmware, but it is packaged separately from the drivers and
from SuSE 9.1 install reports it seems that SuSE just bundles the
drivers and tell their users to get the firmware themselves.
Thanks,
- Michel
19 years, 10 months
suspend & resume
by Rik van Riel
Hi,
I just got suspend/resume to work on my laptop here, by
changing the config files of acpid. I am wondering if
the current acpid configuration (shut the system down
when the power button is pressed) is useful for anyone,
or whether it should be replaced with something else.
One problem I had with the laptop here is that while I
could get suspend to work easily, in order to resume it
I had to press the power button. This meant that after
a resume, the laptop would immediately shut itself down.
If the "shut down when the power button is hit" behaviour
is hurting more people than it helps, I'd like to get it
removed and replaced with something that works ;)
cheers,
Rik
--
"Debugging is twice as hard as writing the code in the first place.
Therefore, if you write the code as cleverly as possible, you are,
by definition, not smart enough to debug it." - Brian W. Kernighan
19 years, 10 months
