scoonma

I know someone out there knows how to do it.

Depends on what you've installed below. With MDV 2008 it's very easy: "urpmi dosbox" as root.

Should I add /usr/lib/kde3 to /etc/ld.so.conf?

That does make sense to me. Afterwards, invoke "ldconfig" (no parameters) as root and try compiling (without --prefix...) --> KMyFirewall should install to /usr/local, which is better in case something fails.

when I try to get an IP address the system complains it cannot reach the DHCP server. Also if I do an ipconfig (Windoze) or ifconfig, I get a couple of addresses which I am not sure where they are coming from... I suppose this is an address obtained from the Linksys wireless router.

Where do you know from? Maybe it's your neighbour...

I'd begin configuring the lan in that order the "internet signal" comes in from the outside, that would as I've understood be:

1. Access point

2. Router

3. PC clients

Is that the correct sequence?

Whenever possible I'd stay away from DHCP. It may be convenient when all works fine, but in case of error tracing and bugfixing it's different.

So at first I'd take a normal ethernet cable and try to reach the access point (if possible) or at least the router and work out a config with fixed IP numbers for each device. Wireless devices would get a single IP if that's the only way to reach them and two IPs if they can be reached both over cable and through air.

This may look like I consider WLAN as an additional option to cable based ethernet connections and not as a basic function - have to admit that applies properly. But it's the only way I'm able to sort this airy(-fairy) stuff out - and it worked in a couple of situations. Best is to fully discard WLAN at home it if possible...

The wineserver I can terminate, but I am left with other running processes called "wine-pthread" which don't want to be killed.

Pthreads are POSIX-Threads:

http://en.wikipedia.org/wiki/Pthread

If a process doesn't like to be terminated in a "civilized" manner, you can do a "kill -9 <process number>" as root. Very few processes can ignore this; for those I happened to encounter I could not find any other solution than terminating the process evoking them in the first place or rebooting.

I believe that you (with help from many other points of view... and that's not a bad thing later when you understand this more) have made this much more complex than it has to be. As I have already noted; the 96xx are not old legacy modules, and will work fine with your card.

Sorry, but I cannot see your point. What does legacy actually mean then? Why should yossarian use a legacy driver when he could use a "normal" one?

Puzzled,

scoonma

What gives?

This is harsh! Now I really know the reason for my decision to still stick to cables...

Don't know what happened, but: What's the meaning of the two modes (client mode / routing mode)? At the router and/or the wireless access point I'd at least enable two different combinations of IP number+hostname for each of your PCs, i.e. same hostname/IP combo for the same PC regardless of operating system.

SuSE or Windows, you'd also have to make sure the other/new IP is visible to the router when (re-)plugging.

If using DHCP beware of possible cross references (maybe decrease default lease time).

HTH,

scoonma

Yossarian,

editing your xorg.conf only makes sense being user "root", no matter which method applied (to be able to use gksu you probably have to "urpmi gksu" before). Of course you can work your way through on this way using gedit, but it's really stony.

As you're obviously trying different options/settings manually (which must not bad in general) I'm advising a different approach:

- Start by disabling the display manager service ("service dm stop" as root)

(- If needed: Relogin as user and "su" to root. )

- Make changes to /etc/X11/xorg.conf using a small terminal editor like joe, nano or with mc.

- Write changes to the file, but let the editor opened with the file still loaded

- Switch to a fresh terminal with "Ctrl-Alt-F2", or "Ctrl-Alt-F3"...

- Login as normal user

- Try your changes by simply typing "startx". If you chose a user login you're used to work with GNOME, the GNOME desktop should load

- Check wether your changes to xorg.conf have been successful

- If you're not satisfied, shutdown the GNOME display and repeat editing in other terminal (hit "Ctrl-Alt-F1" to switch back again)

- When all went fine, logout from your "testing terminal" and (optionally) restart your display manager as root with "service dm restart"

Maybe this sounds a little complicated, but once you get used to this method you'll be able to apply changes much faster than before (no shutdown/reboot necessary with short time feedback to your changes). You'd also need to close X completely to be able to compile your driver anyway.

You don't have to recompile the kernel if you do not *want* to - regardless of your 3D experience. Note the difference? You need the kernel sources to actually build the driver, not the kernel! (cf. dexter's posting from the beginning)

Another word to your driver version: I doubt that you have to rely on the 96xx legacy driver - unless your gfx card is really old. But you can check by yourself on this list of supported products:

http://www.nvidia.com/object/IO_18897.html

If you find your card/gfx chip in here, you can go on with the recent 100.14.19 driver, and you *should* do so. Compiling the that (unfortunately proprietary) driver is easy:

- Download driver file NVIDIA-Linux-x86-100.14.19-pkg1.run

- Shutdown X

- Login as user, "su" to root

- Urpmi kernel sources for your kernel (AussieJohn already gave good advice which file to pick above)

- Enter download location/directory of the driver file

- Enter "sh NNVIDIA-Linux-x86-100.14.19-pkg1.run" and follow the interactive installation

- At the end let the driver file apply changes to /etc/X11/xorg.conf automatically

- Logout (root) and use the new driver

(There's no need to even reboot once.)

Generally, I must admit although considering rpmdrake very useful, it's still ailing when it comes to kernel, dkms and gfx driver issues - which provide basic functions btw. I (partially) know good reasons for the difficulties, but that doesn't help early learners to get all the plenty of fun a linux system of today can provide with acceptable effort.

Regards,

scoonma

Why could this be?

I'm not sure if I understood your config correctly. How many D-Link bridge(s) do you have in use? Only one or one for both computers each? Is there any means to access the status/IP address of these device(s), say by using a webfronted?

Some D-Link bridges offer different modes of work. If you have a device of such type, can you swith them manually, or is it done fully on automatic?

Yossarian, you're running the new kernel 2.6.22.9-1 now, right? It may not be appropriate to to this in conjunction with the old legacy nvidia modules (96xx) - please correct me if your config is different now.

I'd generally adivse either to use the dkms modules for nvida (okay if it works) or download the most recent nvidia driver module from vendor's homepage and install it: Compile the source module by building it against your corresponding kernel sources (best results if it's not too complicated for your liking).

When trying out one of these methods *after* the other, carefully uninstall as many as possible remains from your further attempt. If you decided to go building against kernel source, check /usr/src directory for the symbolic link named "linux" correctly pointing to the subdir matching your kernel name.

After rebooting, the Audigy card worked. Only one stereo channel, but I think that is an unrelated problem.

Nice to hear it's working - basically. :)

You should now be able to adjust channel volumes using "alsamixer" as root. After applying your changes, simply quit the app using <ESC>.

Already have them both as You can see below.

Good. Please inspect the contents of this file:

/etc/ld.so.conf

and the contents of the directory

/etc/ld.so.conf.d/

If there's no KDE library path reference in one of these, you may have to provide that information at compile time using the option mentioned above explicitly (try /usr/lib/kde3 first). Also check wether you've installed all -devel type packages possibly needed for your KDE version. Another potential source of problems I could think of is a possible version conflict or mismatch due to KDE 3.x / 4.x. Mandriva 2008 offers both of them.

Under root - glx: command not found.

I'm pretty sure pindakoe meant the glx entry his xorg.conf is reflecting within the Module section. I'd recommend using it, too.

BTW: Is there any good reason why the option "RenderAccel" is set to "false" in the Device section? I'd set this to true or delete the option line.

Also make sure there's no double entries of options within xorg.conf. Some config program generated these here, located in different sections, but no complaint form Xorg!

You may also need this entry:

Section "Extensions"

Option "Composite"

EndSection

Good luck!

It says Make: command not found.

Refrain from compiling stuff on your own whenever it's possible. In most cases you can install and use preconfigured packages using "urpmi" or a graphical frontend. If there's no other means left, compiling on your own can be helpful - do an "urpmi make" as root to install GNU make. In order to gain a minimal compiling environment, you'd also have to "urpmi gcc" or something similar. You'll probably need that for building a 3D hardware accelerated driver component for your machine. But the less you compile, the fewer headache you will cause for yourself. ;-)

You CAN make a LiveCd with grub or ISOLinux (provided you are a talent at scripting), but AFAIK not LILO.

Why should LILO not be able to do that? I've created a bootable USB stick using LILO with a slax iso. Would doing a CD be so much different? (cf. https://mandrivausers.org/index.php?s=&...st&p=257620 )

Hi Jimbob,

judging from your kernel version you have 2008 running, right? I'm using a customized version of that kernel with an Audigy 4 running fine, so the setup is quite similar.

In /etc/modprobe.conf I've kicked out every sound related entry with no bad consequence; alsa seems to behave more cleanly all in all with the new version. Another detail attracting my attention is this:

00:1f.5 Multimedia audio controller: Intel Corporation 82801EB/ER (ICH5/ICH5R) AC'97 Audio Controller (rev 02)

Do you have an onboard soundcard/soundchip, parallel with your Audigy card? If so, I'd disable the onboard chip completely. I'd try to shut it down using BIOS settings at boot time. If you're lucky, alsa won't see it afterwards, only recognizing your Audigy card so you can (hopefully) simply use alsaconf and you're ready.

Good luck!

Thanks everybody your suggestions solved my problem.

So what did you actually do to get it working again? We're curious. ;-)

Another possibly useful goodie is urpmi-recover. From the docs:

Summary : A tool to manage rpm repackaging and rollback

Description :

urpmi-recover is a tool that enables to set up a policy to keep trace of all

packages that are uninstalled or upgraded on an rpm-based system, and to

perform rollbacks, that is, to revert the system back to a previous state.

Okay, ilia was faster, but here's the long version ;-)

Please check if you still have urpmi installed - so do as root:

"rpm -q urpmi". If the result is positive, do an "urpmi drakconf". This requires you still have software repositories properly installed. If not, you could add them again using easy-urpmi.

If urpmi is gone too, you could download and install the urpmi package manually using rpm.pbone.net or a similar search engine. But this has some possible flaws: First you have to pick the exactly fitting version for your Mandriva system. Besides that you may have to install other packages urpmi depends on which you probably have removed along the way.

Here's a list of requirements for 2007.1, output taken from the server above:

perl(Getopt::Long)	
perl-Locale-gettext => 1.05-4mdv	
perl(POSIX)	
perl(Exporter)	
eject	
/bin/sh	
rpmlib(PayloadFilesHavePrefix) <= 4.0-1	
/usr/bin/perl	
perl(Locale::gettext)	
perl-MDV-Packdrakeng => 1.01	
perl(Cwd)	
gnupg	
gzip	
rpmlib(CompressedFileNames) <= 3.0.4-1	
perl-base	
perl(File::Find)	
perl(MDV::Distribconf)	
webfetch	
perl-URPM => 1.60	
perl-base => 2:5.8.8	
perl(URPM::Resolve)	
perl(URPM)

Depending on what you actually did more or less of these may be missing, too. So if you get a feeling of becoming lost, you may backup your /home directory, if not residing on a partition of it's own and reinstall. A carefully customized install would be proper in the latter case.

If you want to take the high road and gather some valuable experience, try step by step manually first (you'll never erase mcc accidently again after ;-).

Good luck!

Hi there,

the new version of Adobe Reader (AdobeReader_deu-7.0.9-1.i386.rpm) seems to install correctly, but when actually trying to run by typing "acroread <file>.pdf" I got nothing but innumerous messages of this type:

expr: syntax error

Guessing the source of the error was something similar like the one found here, I inspected the file /usr/bin/acroread a little closer. This shell script may be suitable for other rpm based distributions, but not for Mandriva, which seems to be due to a slightly different naming scheme for library packages - at least as far as gtk-libs are concerned. The script tries to find "libgtk-x11" libraries, which simply does not exist for Mandriva. We have "libgtk+-x11" packages instead.

So it is sufficient to replace the "libgtk-x11" strings found in the original script by "libgtk+-x11". That's what I did and afterwards the application started perfectly without complaining.

This method feels like a more adequate solution to the problem itself than simply adding a "0" character into a single sed command line like before; so the method depicted may apply well to earlier Mandriva versions, too.

Any field reports welcome! :)

"a fatal error ocurred: urpmi database locked"

This error occurs when two different install processes try to get access to the database (query is no problem though). In most cases this is indicating an urpmi process which stalled, was cancelled badly or is still running in the background.

In a situation like this I'd either find that task and completely remove it or resume it and stop the latter one.

Did you install the automake package?

EDIT: Please also do a "urpmi libjpeg62" if needed.

On the other hand, if you look at MP3, it really takes a lot of investment to figure out how to do that kind of lossy compression and still end up with something acceptable.

I'm not talking about the implementation or the algorithm (algorithms are not patentable), but it takes a lot of research into psychoacoustics to figure out how to do your filtering and signal discrimination to compress without deteriorating the listening experience.

I'm not sure how one can protect such investments...

We all can be glad MP3 was invented originally at Fraunhofer institute in this country - which gains mixed financing both from business customers *and* public authorities. In consequence of that there's no global patent warfare on it. I'm pretty sure the method of mp3 (de-)coding would not be as widespread if it were patented to the full extent; we had a multitude of similar but incompatible patented techniqes instead - together with lots of converting tools.

I consider research to be more fruitful when interchange of ideas and methods is less restricted. At the very moment big corporate money gets in, quality goes down the drain - because invention becomes an issue of making money, with marketing and the like much more important than product quality by itself. Bugs become features and many users open windows of deception instead of portals of perception now...

If you're clever enough to achieve real innovation you get your first payment in a currency of much higher value than gold coins or pay checks: fame. Money following that does not pose the same social and political problems than we have with software patents.

Why does GPL exist? If there were no patents of software, there wouldn't even be a need to defend the idea of free speech in terms of code.

Well, the 3D stuff is not really necessary. It IS a nice toy and yes, it draws many people to distro A or distro B but in the end, many users still have problems with 3D support while 3D doesn't work at all on many systems. From a productivity point of view, he is right, the 3D desktop is rather a nice gimmick more than anything else but from a marketing point of view it is important to offer it.

I'm slightly disagreeing on the 3D stuff: In the beginning it was sort of a gimmick to me, too. Now I don't want to loose the window selection function offered by compiz; don't know the correct name but it works as follows: If your screen is littered with windows, you can easily choose the desired one by dragging the mouse pointer into the very top right corner. Thus you get an overview with small versions of every opened window on top the desktop and you can easily select by simply clicking on this image, raising that app into foreground. Very handy! (Example attached)

P.P.S. What about menu shades? It is realllllllyyy annoing, and it seems like nobody knows the cure...

Sorry, may sound stupid but: Choose a different theme?

Lookin at it purely mathematically, there'll be a "best way" to perform a task. A single way that will be more optimised than any other routine that does the same thing on the same hardware.

OT: You're right of course, as long as the ideal solution exists (travelling salesman problem,...) and as long as it's a reasonable effort to find it on given hardware. Chess programs and database problems came to my mind - and finally an old movie you probably know, too: Wargames. The optimal solution to the problem was to stop the program from further computing...

It is a widespread idea that mathematics can describe or actually is identical to the underlying essence of nature, of our world. OTOH, how could you code an emotion?