uudruid74

My only comment is that Oracle VirtualBox may not run Linux guests that well, but it works really well with a Win7 guest. I have some test study materials that are Windoze-only (and wine/play-on-linux won't run it in-spite of the fact that the .exe is written in java and needs Sun/Oracle java run-time to run - so much for portability), but the virtual machine even has a seamless mode. I keep my linux dock on the bottom and the Win7 one on the top and both can share the desktop more or less seamlessly. The linux dock only hides when its covered so it stays accessibly and the Windows task-bar only un-hides when you bump the screen edge, so it stays out of my way. I can definately recommend it over a dual-boot unless you need Direct3D or games. Then again, I have 8GB of RAM on my laptop so if you run with less RAM, the performance may not be as good.

OK cool. Will try it tonight (I removed the vala flag for that build to work around it for now). Any ideas on PHP? I've re-emerged the whole system and PHP is the only one that refuses to build :/ Maybe it will be better after the sync. I just don't like syncing that often because I want to stick with a more or less stable set of packages that all build correctly, plus we're bandwith limited here so running out of monthly allotment is an issue.

configure of gedit (with the vala use flag, but manually running ./configure --enable-vala=yes) says ... checking whether to forbid deprecated symbols... no checking for itstool... itstool checking for xmllint... xmllint checking for gobject-introspection... yes checking for pkg-config... (cached) /usr/bin/pkg-config checking pkg-config is at least version 0.9.0... yes configure: error: vapigen >= 0.25.1 not found Taro gedit-3.14.0 # vapigen-0.2 vapigen-0.22 vapigen-0.24 vapigen-0.26 OK, here is the section of ./configure that is failing. Notice all the tests that don't have variables in them. esac if test "x" = "x"; then : vapigen_pkg_name=vapigen else vapigen_pkg_name=vapigen- fi if test "x0.25.1" = "x"; then : vapigen_pkg="$vapigen_pkg_name" else vapigen_pkg="$vapigen_pkg_name >= 0.25.1" fi Is this a bug in the ebuild? If so, how did it get unmasked? Or is there no unit testing to make sure each USE flag works?

Ack! Rochester, NY! I was born there!

Have you see the way grub is configured on the Gentoo liveCD? Doesn't look like grub at all except for the "Welcome to Grub" printed by the EFI loader right before the graphical menu starts. Even boots the kernel in a little window! I have no idea how they pulled that off yet, but I'll be looking into it.

Couple questions since we are on the topic. Will Grub scale the given image to the current resolution, or do you have to provide a copy of the image in every resolution? Is there a tutorial on how to change the menu size? I've seen the menus placed in various parts of the screen, changed to use icons, and all sorts of stuff. The last Gentoo live DVD is a good example. Its a big download, but the grub on it is newer than what funtoo has available and the configuration is very well done! In the loopback mount that you did, have you ever tried to grab the ISO off NFS? I think its possible. And if grub can do that (would need NFS drivers in grub) then you should be able to chainload the ISO image as if it was a CD without having to give it the kernel and initrd lines ... just chainload to whatever secondary bootloader is on the medium. I was thinking that with some careful scripting, you could have an NFS server with a bunch of ISOs and a script that generates a grub config file for all those entries. You would then just need a grub menu item that loads the remote config file. I guess grub can load files through tftp. Anyone tried to load a whole ISO via tftpboot and loopback it in grub?

Take your known "good/stable" kernel image and initrd and copy them to new names that genkernel won't overwrite. The problem with automatically doing this is that you might be running genkernel because the old one doesn't work, which would overwrite the other ".old" that is good, or you end up with a million copies of kernels (like my Ubuntu-based system that likes to upgrade the kernel and break shit all the time and leave with a million kernels in my grub menu.

Probably not the issue, but might be worth trying ... I'm not using Gnome yet, but I do use the ATI drivers (actually using Elementary at the moment, which is based on Gnome). One problem I consistently found is that I would often get odd behavior after certain upgrades. I tracked it down to having an xorg.conf file. Even the one generated by aticonfig causes problems. If I just erase the file, Xorg autodetects everything perfectly - and I have an extra USB wireless keyboard/mouse, a synaptics touchpad, a touchscreen, and a graphics tablet plus an external HDMI monitor (in addition to the laptop's built in screen and keyboard). Thats a lot of devices and they all work fine with no xorg.conf (ok, I actually have a tiny script that tells X that the touchscreen is attached to the built-in screen and not the HDMI - it doesn't know that unless I tell it). You might want to try saving your xorg.conf file so that Xorg doesn't see it and see if that fixes your 3D problem which could in-turn fix gnome. Otherwise .. I have a hammer so this looks like a nail :) I don't know what else you could try.

I wish it was just a USE flag. I used equery to find out what the currently installed php USE flags were. I tried emerging it with the original flags (adding them to package.use) and it still fails with the sandbox violation. I could keep this setting in there so that the system doesn't try to recompile it, but ... 1 - I want to get rid of the dependencies for those flags, 2 - this tells me something is broken and I need to fix it before something else breaks. In the meantime, I have no idea what to do about this. I don't understand why php is looking at the cgroup stuff or why portage blocked it from doing so. Surely configure scripts are allowed to look into /sys! At least ... it worked before. I'm recompiling everything hoping that something will reinstall that fixes it. I need PHP for cacti :)

I installed everything that way without problems. Not realizing that the defaults for quickpkg don't save config files ... but using the packages in an emerge will write blank ones over my /etc files ... I have a nicely upgraded system, but once the emerge stopped, its all broken. I've done the best I can to repair my /etc files, but something is stopping this package from installing. I did try setting /bin/sh to bash just to be sure that wouldn't fix it. No dice. Any other ideas and what it could be?

OK - I went to change a few USE flags (turning off some stuff I don't need like LDAP), and PHP won't emerge. Its the same version. I don't know how I broke it :/ The errors appear to be access-denied errors to /sys/fs/cgroup config.status: creating ext/phar/phar.1 config.status: creating ext/phar/phar.phar.1 config.status: creating main/php_config.h config.status: executing libtool commands config.status: executing default commands >>> Source configured. * --------------------------- ACCESS VIOLATION SUMMARY --------------------------- * LOG FILE: "/var/log/sandbox/sandbox-17465.log" * VERSION 1.0 FORMAT: F - Function called FORMAT: S - Access Status FORMAT: P - Path as passed to function FORMAT: A - Absolute Path (not canonical) FORMAT: R - Canonical Path FORMAT: C - Command Line F: open_wr S: deny P: /sys/fs/cgroup/test.shm.27131JlCbSd A: /sys/fs/cgroup/test.shm.27131JlCbSd R: /sys/fs/cgroup/test.shm.27131JlCbSd C: ./conftest F: open_wr S: deny P: /sys/fs/cgroup/test.shm.104439yZW3n A: /sys/fs/cgroup/test.shm.104439yZW3n R: /sys/fs/cgroup/test.shm.104439yZW3n C: ./conftest * -------------------------------------------------------------------------------- >>> Failed to emerge dev-lang/php-5.5.18, Log file: >>> '/var/tmp/portage/dev-lang/php-5.5.18/temp/build.log'

Well - not currently running Funtoo while I clean up a mess and get everything tweaked the way I want it. This is my old system - a combination of Elementary Beta with added UbuntuStudio packages and the Cairo-Dock instead of a desktop. My terminal slides down from the top with a hot-key. You get brownie points if you can guess whats on the second screen. I dropped the jpg quality way down because its so big.

Well, I decided that after changing various flags and upgrading my C compiler, I would go ahead and recompile my whole system to make sure that everything compiled cleanly and to make sure there would be no breakage in the future. I started with an emerge gcc to get the new compiler. Then I did emerge -e @system to recompile all of the system-level stuff with the new compiler. Good so far. Next I went to do -e @world, but to make sure it wouldn't do @system again, I decided to just quickpkg @system and then let emerge install from those packages if it hit a dependency. Sounds logical right? Little did I know that quickpkg has a default setting that does not save configuration files. It stores blank ones with a note. Then when world installed them ... well, it wiped out a lot of my /etc files. YIKES! And to make it worse, I figured I could rescue the vast majority of them by pulling the originals from the old stage3 and let it overwrite my new blank ones ... which broke all sorts of stuff because /etc/ld.so.conf was pointing to the wrong gcc libs because I had just upgraded it. *sigh* I'm still not 100% sure my carefully tweaked make.conf is right, but it should be close (I had a fairly recent backup). So .. don't do that!

The sliding penguins are definately cool, although it doesn't convey a single "mascot" to rally behind. Here is the closest I could find on Tux Factory: http://tux.crystalxp.net/en.id.2358-snow-boy-v2.html Although I didn't spend a lot of time searching. Can anyone compare/contract fbsplash with Plymouth? Is plymouth safe to use with hardened? I know fbsplash will keep the console pretty after boot finishes and plymouth doesn't, but if it won't work on hardened I think its kinda pointless. The way I look at it, most people using desktops will never see the consoles. Most people running servers will probably find the graphics distracting or are running hardened and can't run it. I think the people that want a fancy console will be the minority (and those guy can probably install fbsplash themselves). That said, gentoo has info on how to install either one. Is genkernel-next required to use plymouth? I'm not sure what benefits genkernel-next gives me and I installed genkernel already. Would I need to switch to install plymouth?

There is a different between EFI and UEFI. The latter has signed security certificates, which I don't think funtoo supports (not sure) since someone has to be the certificate signer. This would be a bad combination as you would have to sign your kernel ever time you make a change and then manually tell the BIOS to accept the new certificate. Its a little bit easier to just have a signed GRUB and I think GRUB will bypass the rest of the "secure boot" options. If you can turn off the "secure boot" options in your BIOS, start there. Other than that, make sure you have an EFI partition that is fat or vfat formatted. Most EFI systems will let you choose an EFI file to boot. You can grab one of the shell efi files and boot that for testing. I always keep a couple different EFI files around (including an efi version of memtest, etc). Keep me posted. I'm more familiar with grub than EFI stub, but I canget around EFI fairly well these days. There is a learning curve, but you get a couple extra options if you care to take advantage of them.

UHmm ... 'forced' to maintain init scripts? Dude. Hate to tell you this, but this is UNIX, not Windows. If you develop a Unix service, you can write the init script or let the admin do it, but this has been a standard for ... 40 years? Thats like saying a Windows developer is "foced" to have an installer program. A shell-less boot? I hate to tell you this, but a shell is part of UNIX. I need a shell no matter what. I don't need DBUS. A UNIX shell is a requirement of a UNIX system. You are adding requirements, not taking them away. Linux = Linus + Unix Just proof that systemd mongers are trying to take the Unix (and the FREEDOM of Unix - I LIKE to modify my scripts, thank you) out of my Linux box and replace it with a bunch of junk. I was really trying to give the systemd crowd the benefit of the doubt, but you went off the deep-end with the worst arguments I've ever heard. You are definately troll-bait! Adding to my block-list

I don't mind people using systemd at all. Its the lock-in that people are being forced to use it that I don't like! And starting services with sockets and dbus just seems like a BAD idea to me. I want as little as possible in-between my services and getting them up and running - the less to break the better. I don't like adding MORE single points of failure. I'm sure its fine on a desktop, but not if I'm forced into it. I think systemd had good intentions, I can see what they were trying to do, but I don't like the way they want about it, nor do I like Gnome being so reliant on it (or any other service), but considering whos doing the backing on these projects ...

Hmm ... thank you for all the information. Its quite interesting. Would it be possible to patch ConsoleKit and pm-utils so that Gnome can use the same API that it uses to talk to systemd's utilities? Sort of pretending to be systemd? Or would that not be possible? I'm just thinking that if you have to fork and maintain other utilities, perhaps it would be possible to avoid patching gnome as well (or waiting for them to put back the support upstream now that everyone is jumping on the systemd bandwagon). I'm not sure what methods systemd uses, but I suppose if its a single socket for both services, expecting a "systemd" to do it all, then this makes it almost impossible to replace without modifying Gnome. And I'm guessing that is the problem. I'm surprised there isn't some sort of directory service API you could talk to on start-up (maybe in dbus) that says "who do I talk to for power management" ? or "who do I get session service information from"? And then machines without systemd could be pointed to the specific daemon for that service while machines based on systemd would just be told "systemd" for every answer. Or am I missing part of the picture?

And I heard that Gnome itself didn't depend on systemd, but the login manager depended on logind, which was in-term systemd. Was there more to it?

I have two questions: Whats the point of wayland? From what I can tell, its supposed to be faster, but is it really? Won't we be losing the network transparency of X? I kinda like being able to SSH into a host and run a GUI app. And I don't think VNC is an alternative. VNC still doesn't work as well or as fast as remote X. I'm in no hurry to give up X. And I hear games are actually slower, and that is where speed matters most (I don't play games much, but thats not the point). A few things I do with X: I have a script that ties into bash that remembers what the window ID is for the window its running in. When a command completes, it checks the top-window and compares it. If its not currently the top window, it displays a notification of the command that just completed, the window's name, and its icon. The window title is always updated with the current title, dropping back to just the shell name when the command completes - unless its not the top window at command completion - in that case it leaves the last command in the window title (and icon name if its minimized) to add in finding the window (in case you have multiple windows plastered all over the screen). Would this sort of hackery (from a plain bash shell script) be possible with Wayland? And this works remotely too - it's a bit harder to get the notify working remotely because of DBUS (although I could use 'dialog' to provide an alert and it would be easy and portable - just not as pretty, and so I'm forced into using backwards-ass dbus). I have another app called X2X that lets me use a single keyboard and mouse with multiple machines - you move the cursor off one screen and it moves to another computer (over the network) via remote X protocol. I don't think Wayland can do that! @Dantrell: Great work on Gnome! Just curious, you said "nearly complete". What parts had to be removed? As for the rest of the topic, only 12% (gentoo forum vote) think gentoo should use systemd. I understand keeping it around in portage as an unsupported option (vs Gentoo's supported option), but I don't see either distro changing from OpenRC as the default. Not for 12%! I wonder how many people on other distros actually think it was a good move? Personally I'm in the anti-systemd camp myself and I think more people would use alternatives if better ones existed. Yes, funtoo is great, but the rather long compile and config process is a turn off for many, and once you are done, you may not get all the bells and whistles that other distros have without more tweaking and recompiling. The systemd (and kdbus) issue is becoming a rather heated debate, almost as bad as Atari vs Amiga or Vi vs Emacs, and its really silly. Everyone knows vi is better!

l33t- do you really expect anyone to take you seriously with that name? I'm really curious just how much faster its really booting - are we talking 2 or 3 seconds, or 10 or 15? And how many services are simply delayed to continue start-up after gdm has loaded? Perhaps a better test would be to have the system boot straight to the desktop without a login. Other questions: 1 - Have you compared to OpenRC's parallel option? 2 - Have you tried changing /bin/sh to point to dash instead of bash? This will make your scripts run faster as long as they don't require bash extensions. I know a couple or packages seem to expect /bin/sh to have bash extensions when you emerge them, so expect problems. I hope to have some real information on how these two affect performance and if they are worth it very soon now. The Gentoo flame war is only up to 30 pages, not 46

Oh yeah, I don't mean manually changing the title. I mean having emerge show what its emerging automatically in the title. I actually edited it (slight bug if package name ends in an r or a). Portage/Emerge also does this, but uses the wrong escape sequence for the "screen" terminal type. Its done through an ANSI escape sequence, the same way it does colors, but in this case its not portable to all terminal types. I actually plan on making another change. Both Compiling and Cleaning start with the same letter, so just having C isn't descriptive and I don't have enough room for the whole word. I'll have to make the hack a little more involved. Speaking of neat hacks. Open a terminal and type emerge --moo I'm sure a lot of people have seen it, but perhaps you haven't. Little Easter Egg. My terminal set-up: I'm actually using a dual layered approach. I use byobu which is a layer over tmux which is based on screen. This is a standard text application, not graphical. Yeah, sounds complicated, but this has a number of features: It can multiplex your terminal, so that you can have multiple shells running on the same screen. Yes you can do this with your GUI terminal, but this works even over SSH and can do split screens and such. There is a lot of flexibility - you can even cut-n-paste without a mouse (mouse is still optional). It does session management. If I log in to SSH, I get the same shells that are on my desktop. I can even log in with my phone and the byobu terminal size will change as I rotate the phone. Session management means I can log in remotely (by phone) to check on the status of a long compile, or log out of my desktop and log in to a new one ... and my shells are uninterrupted - compile still running. The status line tells me system information such as how hot my CPU is, CPU speed, etc. For some odd reason, the termcap entry is "screen", not "xterm" And since I do a lot of work in the shell I need access to it instantly, but I don't want to open a new window since I really only need a single window. Byobu is multiplexing the contents of the window, so I don't need tabs. In fact, most terminals have a tendency to take up too much space and its easy to lose a couple of them. Enter tilda: Tilda is a pop-up terminal. I configure it to get rid of all the things I don't use. No tabs, no title, no scrollbar. You hit a hot key and it appears (all workspaces). Hit the key again, and its gone. And you can configure what "shell" it runs, so I have it running byobu. ssh runs a "login shell", so it gets byobu. All other terminals (if I run on) are normal bash (bypassing byobu).

Do very many people here create ebuilds? After all, most everything that I can think of already has an ebuild or an overlay. I suppose if the process to create your own ebuilds were a bit more sane, we could easily do the same with funtoo using the forums.There are votes and file attaches here right? Right now, the only ebuild I need is actually porting the catalyst-test from arch's AUR. They have things working with the latest Xorg and released kernel which doesn't seem to have a Funtoo/Gentoo equivalent yet. And while I don't play games, I found LyX to be horribly slow (unusable - it will confuse X and overflow the input buffers) with the open-source drivers. Something about the binary drivers works around the horrible display code in LyX, and I use LyX a lot, and I've found that the newer the ati-drivers, the fewer problems I have. All the binary distros want to use versions that are ancient and then complain about the bugs. Well ... duh! I know what you mean about the long compiles - I don't mind once the system is installed since I can use it while it compiles. Its the install that gets me. And my new hack tells me when a command completes in a terminal window that is hidden, so I get a notify when the emerge is done.

Actually C++. Interesting that its slower! I would expect the opposite - python isn't known for speed. As for his claims on readability of the code, I'm not a huge python fan, but I have nothing against it. C++ I really don't like. And reading through the site, I kind of got the impression that the developer was another LP-type, a bit too full of himself it seemed. Thanks for the verification on this. That makes a bigger difference over a script running faster or slower by a few seconds, especially when the compile itself will be taking up the largest percentage of the time rather than the package manager.

OK - my mistake. Apparently, this is supposed to be automatic, but its not working with my setup. My TERM is "screen" which is supposed to work, but the terminal is running tmux. I assume screen and tmux are compatible, but setting the window title can be done two different ways. There is also the issue that window titles should be kept small when running a terminal multiplexer, so I hacked the right escape sequence in and added a regex to shorten the information (turned "Compile" ->C, "1 of 2" -> "1/2", removed the version from the package, and chopped up the rest). If anyone else finds this useful, modify /usr/lib/portage/pym/portage/output.py starting at line 257: if not raw : if os.environ.get('TERM') != "screen": mystr = '\x1b]0;%s\x07' % mystr else: mystr = re.sub(r'.*?\((\d+) of (\d+)\).*?/(.*?)[-0-9\.ra]+\s(\w).*', r'\4-\1/\2 \3', mystr) mystr = '\033k%s\033\\' % mystr # avoid potential UnicodeEncodeError I'm not terribly familiar with python, so that could be a horrible hack, probably is, but it works. Now works in all my terminals, with very small "tabs" in my tmux/byobu screen.