Re: [gentoo-user] Binary package server questions
On Tue, Feb 21, 2017 at 12:22:51AM +, Neil Bothwick wrote > If the chroot is identical to your netbooks's install in terms of > *FLAGS, USE, @world etc, then yes. I used to do it this way when I had an > Atom netbook. I even build for a low memory 486 system in the same way. Unfortunately, the cpus are different enough that CFLAGS, CXXFLAGS, and CPU_FLAGS_X86 are different. See the web page... https://gcc.gnu.org/onlinedocs/gcc-4.9.4/gcc/i386-and-x86-64-Options.html#i386-and-x86-64-Options I actually use "-march=native" in make.conf, but this translates as... * The netbook cpu equals "-march=bonnell" (first-generation Atom). * My current desktop equals "-march=ivybridge" * My "hot backup" machine equals "-march=silvermont" The natively-compiled code on the netbook will not run on my desktop because the Ivybridge cpu doesn't support MOVBE instructions. The netbook has 2 gigs of ram. I estimate 12 hours if I * changed CFLAGS to "-march=core2" and adjusted CPU_FLAGS_X86 * and ran "emerge -e @world" on the netbook. That would produce "lowest common denominator" code that would run on both the netbook and my desktop. Then I could rsync the contents of the netbook into what would become a chroot directory on the desktop. After that, I'd have to change to "-march=native", adjust CPU_FLAGS_X86, and then "emerge -e @world" on both the netbook and the desktop's chroot. A better option would be to * rsync the contents of the netbook to the Silvermont. It's a newer Atom-family cpu, and can handle MOVBE instructions. * change CFLAGS to "-march=silvermont -mno-movbe" and "emerge -e @world" on the Silvermont. It has a newer, more powerful cpu than netbook, and also 8 gigs of ram, versus the netbook's 2 gigs. A full rebuild won't take anywhere near as long. * rsync the contents of the Silvermont's chroot directory to the Ivybridge desktop. > Oh, and you don't need a package server, just export PKGDIR via NFS > and mount it on the netbook. I see nfs as being more complex with kernel settings required for client and server, not to mention config files all over the place. Gentoo has python as part of the system. To fire up a very simple binary package webserver from a commandline (xterm/whatever)... In python 2.x cd /usr/portage/packages python -m SimpleHTTPServer In python 3.x cd /usr/portage/packages python3 -m http.server ...where "" is the desired port number to listen on. In both cases the default port is 8000 if not specified. Note that only root can open privileged ports in the range 0..1023. -- Walter DnesI don't run "desktop environments"; I run useful applications
Re: [gentoo-user] Binary package server questions
On Tuesday 21 Feb 2017 00:22:51 Neil Bothwick wrote: > On Mon, 20 Feb 2017 18:34:47 -0500, Walter Dnes wrote: > > Reading https://wiki.gentoo.org/wiki/Binary_package_guide still leaves > > > > me uncertain. I have an ancient 32-bit Atom netbook. I've installed > > uclibc-ng Gentoo on it. Building big packages on it is a pain. I can > > do an identical install in a QEMU VM, and distcc into it. But that > > doesn't catch all compiling work. > > > > What I'd like to do is build binaries in a chroot on my desktop, > > > > assuming a 32-bit uclibc-ng chroot on a 64-bit glibc host is possible. > > Because the cpus are different, I would need to use different CFLAGS > > (and CXXFLAGS) variables for when the host updates its own files, versus > > when it builds files for the netbook. > > If the chroot is identical to your netbooks's install in terms of > *FLAGS, USE, @world etc, then yes. I used to do it this way when I had an > Atom netbook. I even build for a low memory 486 system in the same way. You'll need to run in 32bit mode when chrooting of course: linux32 chroot /mnt/Atom_Build_env /bin/bash source /etc/profile export PS1="(Atom_Build) $PS1" > > Finally, is it possible for the client (the netbook) to notify the > > > > host that it needs certain packages built? I plan to run with > > "--getbinpkgonly" on the netbook. > > You don't need to if the systems are the same. Set both systems to use > the same $PKGDIR, set FEATURES=buildpkg in the chroot and do a world > update. Then do the same update on the netbook but with -K. > > I used a script to control this that basically synced world and most > of /etc/portage before entering the chroot, although I later switched to > using containers as they make life so much easier. > > Oh, and you don't need a package server, just export PKGDIR via NFS and > mount it on the netbook. Or, if you can't be bothered with the extra work to set up NFS, copy the contents of the PKGDIR from the chroot'ed system to the Atom after you finished building all the chroot'ed binary packages, then emerge world in the Atom. -- Regards, Mick signature.asc Description: This is a digitally signed message part.
Re: [gentoo-user] Binary package server questions
On Mon, 20 Feb 2017 18:34:47 -0500, Walter Dnes wrote: > Reading https://wiki.gentoo.org/wiki/Binary_package_guide still leaves > me uncertain. I have an ancient 32-bit Atom netbook. I've installed > uclibc-ng Gentoo on it. Building big packages on it is a pain. I can > do an identical install in a QEMU VM, and distcc into it. But that > doesn't catch all compiling work. > > What I'd like to do is build binaries in a chroot on my desktop, > assuming a 32-bit uclibc-ng chroot on a 64-bit glibc host is possible. > Because the cpus are different, I would need to use different CFLAGS > (and CXXFLAGS) variables for when the host updates its own files, versus > when it builds files for the netbook. If the chroot is identical to your netbooks's install in terms of *FLAGS, USE, @world etc, then yes. I used to do it this way when I had an Atom netbook. I even build for a low memory 486 system in the same way. > Finally, is it possible for the client (the netbook) to notify the > host that it needs certain packages built? I plan to run with > "--getbinpkgonly" on the netbook. You don't need to if the systems are the same. Set both systems to use the same $PKGDIR, set FEATURES=buildpkg in the chroot and do a world update. Then do the same update on the netbook but with -K. I used a script to control this that basically synced world and most of /etc/portage before entering the chroot, although I later switched to using containers as they make life so much easier. Oh, and you don't need a package server, just export PKGDIR via NFS and mount it on the netbook. -- Neil Bothwick Artificial Intelligence usually beats real stupidity. pgpA1Z4f_R6U4.pgp Description: OpenPGP digital signature
[gentoo-user] Binary package server questions
Reading https://wiki.gentoo.org/wiki/Binary_package_guide still leaves me uncertain. I have an ancient 32-bit Atom netbook. I've installed uclibc-ng Gentoo on it. Building big packages on it is a pain. I can do an identical install in a QEMU VM, and distcc into it. But that doesn't catch all compiling work. What I'd like to do is build binaries in a chroot on my desktop, assuming a 32-bit uclibc-ng chroot on a 64-bit glibc host is possible. Because the cpus are different, I would need to use different CFLAGS (and CXXFLAGS) variables for when the host updates its own files, versus when it builds files for the netbook. Finally, is it possible for the client (the netbook) to notify the host that it needs certain packages built? I plan to run with "--getbinpkgonly" on the netbook. -- Walter DnesI don't run "desktop environments"; I run useful applications
Re: [gentoo-user] Re: How to dump kde gracefully in favor of lxde
On Mon, Feb 20, 2017 at 01:44:02PM +0100, Kai Krakow wrote > You could also try to stop portage from even syncing the KDE components > into the tree. I usually do this for small systems to not give portage > any chance of pulling in unwanted components. As a plus, syncing and > dep calculation should be faster. That would generate error messages. The best solution is to get rid whatever is pulling in KDE. As for blocking stuff to pull in, to quote Frank Sinatra, "I did it m-y-y-y-y-y way". I wrote a script, /etc/portage/cleanup which generates /etc/portage/rsync_excludes on my machines. Note that the groups listed are what I don't use. Your machine(s) will probably have a different set of unnecessary stuff. #!/bin/bash remove() { rm -rf /usr/portage/${1}/ /usr/portage/metadata/md5-cache/${1}/ echo "${1}/" >> /etc/portage/rsync_excludes echo "metadata/md5-cache/${1}/" >> /etc/portage/rsync_excludes } # # Remove rsync_excludes rm /etc/portage/rsync_excludes remove app-emacs remove app-leechcraft remove app-mobilephone remove app-pda remove app-xemacs remove dev-dotnet remove dev-embedded remove dev-haskell remove dev-java remove dev-qt remove dev-ros remove dev-ruby remove java-virtuals remove kde-apps remove kde-base remove kde-frameworks remove kde-misc remove kde-plasma remove lxde-base remove lxqt-base remove mate-base remove mate-extra remove net-p2p remove ros-meta remove sec-policy remove www-apache remove xfce-base remove xfce-extra -- Walter DnesI don't run "desktop environments"; I run useful applications
Re: [gentoo-user] /etc/portage/profile vs. /etc/portage/make.profile
On Mon, 20 Feb 2017 19:15:57 +0100, meino.cra...@gmx.de wrote: > ...I was irritated bu the header line of the linked article and the > fact, that /etc/portage/profile didn't exist in my installation. > I added the directory by hand and hope it is all ok now... You'll also find /etc/portage/package.use doesn't exist, nor any of the other optional directories or files in /etc/portage. Just create them as you need them. -- Neil Bothwick TERROR: A female Klingon with PMS. pgpY71kFf9lxJ.pgp Description: OpenPGP digital signature
Re: [gentoo-user] rdate "time.nist.gov" && hwclock --systohc
It should be: usr/bin/rdate -s "time.nist.gov" && hwclock --systohc Thelma On 02/20/2017 12:45 PM, the...@sys-concept.com wrote: > On one of my boxes system date is 2-days behind. > When I run: > usr/bin/rdate "time.nist.gov" && hwclock --systohc > rdate: [time.nist.gov]Mon Feb 20 12:39:59 2017 > > eden ~ # date > Sat Feb 18 19:28:39 MST 2017 > > The system clock is not getting updated, why? Is the time difference too > much? >
[gentoo-user] rdate "time.nist.gov" && hwclock --systohc
On one of my boxes system date is 2-days behind. When I run: usr/bin/rdate "time.nist.gov" && hwclock --systohc rdate: [time.nist.gov] Mon Feb 20 12:39:59 2017 eden ~ # date Sat Feb 18 19:28:39 MST 2017 The system clock is not getting updated, why? Is the time difference too much? -- Thelma
Re: [gentoo-user] Re: How to dump kde gracefully in favor of lxde
On Monday 20 Feb 2017 15:32:25 Grant Edwards wrote: > On 2017-02-19, Mickwrote: > >> And what pulls in NetworkManager? KDE's power manager with USE=wireless! > > I despise NetworkManager. Over the years, it has been the cause of > countless problems and hours of wasted time. The first I do when > dealing with network problems on *buntu systems is uninstall > NetworkManager. > > > Yes! Madness. What's wrong with good ol' wpa_supplicant and its GUI? > > Which is spelled "emacs /etc/wpa_supplicant.conf I mentioned the GUI in case there were people who preferred it. The CLI equivalent is wpa_cli of course. ;-) -- Regards, Mick signature.asc Description: This is a digitally signed message part.
Re: [gentoo-user] Re: package.provided syntax for overlay
On Monday 20 Feb 2017 13:23:15 Neil Bothwick wrote: > On Mon, 20 Feb 2017 14:07:39 +0100, Kai Krakow wrote: > > > > Another option is to copy/symlink the specific package you want > > > > from the bar overlay to your local overlay and do not include the > > > > bar overlay in repos.conf. > > > > > > Sorry for being dense. Do you mean first add the overlay with > > > 'layman -a bar', then symlink the particular package to my local > > > overlay? How will I be updating this package in the future, if I do > > > not have the 'bar' overlay settings > > > in /etc/portage/repos.conf/layman.conf? > > > > > > I'm trying to understand the benefit of doing it as you suggest > > > above ... :-/ > > > > You could package.mask */*::bar and then only unmask the needed bits. > > That does seem a cleaner way of doing it. When I first did this, that > option wasn't available. Thank you both, eventually I used Kai's suggestion, rather than having to mask a package at a time that entrance was pulling in as dependencies. Entrance works fine, other than the pam config it ships with which is not working at all on a non-gnome gentoo system. I need to brush up on pam, which I have been avoiding it seems forever and this may be the excuse I needed to do it. -- Regards, Mick signature.asc Description: This is a digitally signed message part.
Re: [gentoo-user] /etc/portage/profile vs. /etc/portage/make.profile
Johannes Rosenberger[17-02-20 17:06]: > On 20.02.2017 03:37, meino.cra...@gmx.de wrote: > > Hi, > > > > I am a little confused... > > > > In search for the reason my new root has no /etc/portage/profile > > but an /etc/portage/make.profile on this documemnt: > > https://dev.gentoo.org/~zmedico/portage/doc/man/portage.5.html > > I found this ywo lines > > > > /etc/portage/make.profile/ or /etc/make.profile/ > > site-specific overrides go in /etc/portage/profile/ > > > > (My profile is set correctly) > > Does this mean that make.profile is totally identical to > > make.profile, but if you want site-specific overrides you need > > to have profile instead of make.profile? > > > > Do I need to create /etc/portage/profile if it is not there? > > > > Nonetheless the listing below implies, that they are equivaltent. > > Or ? > > > > Cheers > > Meino > make.profile is a symlink to your profile, so you shouldn't edit > anything in there. It is managed by 'eselect profile' and i guess that > /etc/make.profile is the old location, like with make.conf which used to > be in /etc but is now in /etc/portage. > If you want to manually override anything in your profile that is not > covered by the other files in /etc/portage, e.g. unmask useflags, then > you use /etc/portage/profile. Everything there has higher precedence > than things in make.profile. > Hi Johannes, ...I was irritated bu the header line of the linked article and the fact, that /etc/portage/profile didn't exist in my installation. I added the directory by hand and hope it is all ok now... Thanks for your posting! :) Cheers Meino
[gentoo-user] Is this a dependency bug?
I installed weasyprint-0.29, but it won't run: $ weasyprint Traceback (most recent call last): File "/usr/lib/python-exec/python2.7/weasyprint", line 6, in from pkg_resources import load_entry_point [...] File "/usr/lib64/python2.7/site-packages/pkg_resources/__init__.py", line 849, in resolve raise DistributionNotFound(req, requirers) pkg_resources.DistributionNotFound: The 'CairoSVG<2,>=1.0.20' distribution was not found and is required by WeasyPrint I have cairosvg installed, but apparently it's not recent enough (1.07 vs. 1.20)? $ emerge --search cairosvg * media-gfx/cairosvg Latest version available: 1.0.7 Latest version installed: 1.0.7 Size of files: 29 KiB Homepage: http://cairosvg.org/ Description: A simple cairo based SVG converter with support for PDF, PostScript and PNG formats License: LGPL-3 Is this a dependency bug in the weasyprint ebuild? -- Grant Edwards grant.b.edwardsYow! NANCY!! Why is at everything RED?! gmail.com
Re: [gentoo-user] Gentoo on a Surface Pro 3?
Well even if you can't figure it out, I'd be happy to do things manually if you open an issue on github for me with some instructions. I know from experience what a pain in the ass git can be for first-timers ;-) On 20/02/17 15:33, Daniel Frey wrote: On 02/19/2017 09:40 AM, Daniel Quinn wrote: On 30/01/17 08:24 PM, Daniel Frey wrote: I acquired (on the cheap) a used Surface Pro 3 with the keyboard cover off of a relative who wasn't using it (they said the screen was too small.) I am considering putting Gentoo (or attempting to) and am wondering if anyone has had success. It looks like newer kernels have some builtin support for the hardware. Due to its form factor I will be setting up distcc to help with the build process, and using -bin packages for monstrosities like firefox and libreoffice. I'm sorry that I didn't see this sooner, but I've been running on my SP3 now for about 2years. I maintain a GitHub repo with kernel configs and a installation manual, so if you'd like to give that a spin and submit patches/ideas as they come, I'd be happy to merge them: https://github.com/danielquinn/Gentoo-Surface-Pro-3 Thanks, After messing around manually with the kernel I found your .config, that could've saved me a lot of time. I have a .config for 4.9.6-r1 I believe. I've noticed some patches have been partially applied in 4.9.6-r1 and will modify the patches accordingly. If I can figure out how to use git I can send the updates. Dan
[gentoo-user] Re: How to dump kde gracefully in favor of lxde
"Walter Dnes"writes: [...] > 1) "eselect profile list" and switch to a basic non-KDE profile of your > choice. Moved from: default/linux/amd64/13.0/desktop * to default/linux/amd64/13.0 * > > 2) "emerge gentoolkit" if not already present. > > 3) "cat /var/lib/portage/world" and see what KDE stuff you have. No kde in there > 4) Unmerge (i.e. "emerge --unmerge) obvious KDE-related stuff that you > find in world. got that accomplished and and several more pkgs that were causing blocks > 5) "emerge --depclean" (May not help if you've done "emerge --sync" and > not fully updated). > Nothing gets listed there >The next 3 steps are going to be repeated several times > > 6) "emerge -pv --changed-use --deep --update @world" > > 7) You'll probably see portage try to pull KDE back in. For each lib > "fu-bar/foobar" that portage tries to pull in do "equery d fu-bar/foobar" > and manually unmerge whatever it finds. (Note: gentoolkit provides the > equery tool). > 8) GOTO 6 (until portage stops trying to pull in KDE stuff). I'm still working on this, but wanted to thank you for your input. I like help that has an outline of how to go at something. Very helpful.. and I seem to be close to getting this cleaned up.
Re: [gentoo-user] Gentoo on a Surface Pro 3?
On 02/19/2017 09:40 AM, Daniel Quinn wrote: > On 30/01/17 08:24 PM, Daniel Frey wrote: >> I acquired (on the cheap) a used Surface Pro 3 with the keyboard cover >> off of a relative who wasn't using it (they said the screen was too > small.) >> >> I am considering putting Gentoo (or attempting to) and am wondering if >> anyone has had success. >> >> It looks like newer kernels have some builtin support for the hardware. >> Due to its form factor I will be setting up distcc to help with the >> build process, and using -bin packages for monstrosities like firefox >> and libreoffice. > > > I'm sorry that I didn't see this sooner, but I've been running on my SP3 > now for about 2years. I maintain a GitHub repo with kernel configs and > a installation manual, so if you'd like to give that a spin and submit > patches/ideas as they come, I'd be happy to merge them: > > https://github.com/danielquinn/Gentoo-Surface-Pro-3 > Thanks, After messing around manually with the kernel I found your .config, that could've saved me a lot of time. I have a .config for 4.9.6-r1 I believe. I've noticed some patches have been partially applied in 4.9.6-r1 and will modify the patches accordingly. If I can figure out how to use git I can send the updates. Dan
[gentoo-user] Re: How to dump kde gracefully in favor of lxde
On 2017-02-19, Mickwrote: >> And what pulls in NetworkManager? KDE's power manager with USE=wireless! I despise NetworkManager. Over the years, it has been the cause of countless problems and hours of wasted time. The first I do when dealing with network problems on *buntu systems is uninstall NetworkManager. > Yes! Madness. What's wrong with good ol' wpa_supplicant and its GUI? Which is spelled "emacs /etc/wpa_supplicant.conf -- Grant Edwards grant.b.edwardsYow! My haircut is totally at traditional! gmail.com
Re: [gentoo-user] How to keep my system from (nearly) freezing?
On 02/19/2017 07:53 AM, Helmut Jarausch wrote: > Hi, > > sometime I have some memory hungry ebuilds in the background, when I > start (e.g.) Chromium which needs very much memory if you have a lot of > open tabs. > > In that case my system nearly freezes. I cannot even kill chrome. > What can I do in that case. (Remote login doesn't work either) > > Can I have any additional program (like Chromium) die if there is not > enough memory. Yes, just disable swap as that is what happens when you run out of swap. You can configure the likelihood of a process being killed through /proc//oom_adj. If you disable it for all processes after boot (ie. through a script) then only the "additional" programs can be killed but that may any program that's using a lot of memory so it may not be a good idea for a desktop. See http://www.oracle.com/technetwork/articles/servers-storage-dev/oom-killer-1911807.html > Many thanks for a hint, > Helmut > -- Fernando Rodriguez
[gentoo-user] Re: package.provided?
Am Tue, 14 Feb 2017 23:40:37 + schrieb Neil Bothwick: > On Tue, 14 Feb 2017 23:00:03 +0100, Johannes Rosenberger wrote: > > > I find the package.*-dirs very nice, too. Unfortunately, the tools > > like emerge, flaggie etc. seem to not always use the same file to > > write to, so the files get messed up over time. > > Portage always writes to the end of the last file in the directory, to > make sure that its entry is not overridden by a subsequent entry. I > create an empty file called zzz-auto-unmask in package.use etc. Then I > can move the settings from that file to a more suitable place. This is also what I did, with exactly the same name. :-) -- Regards, Kai Replies to list-only preferred. pgpmopalUe9ka.pgp Description: Digitale Signatur von OpenPGP
Re: [gentoo-user] How to keep my system from (nearly) freezing?
Hi, On Sun, 19 Feb 2017 13:53:49 +0100 Helmut Jarausch wrote: > Hi, > > sometime I have some memory hungry ebuilds in the background, when I > start (e.g.) Chromium which needs very much memory if you have a lot of > open tabs. > > In that case my system nearly freezes. I cannot even kill chrome. > What can I do in that case. (Remote login doesn't work either) > > Can I have any additional program (like Chromium) die if there is not > enough memory. 1. Use reasonable -j and -l options in MAKEOPTS. A good start will be -j N and -l 2*N, where N is a number of your logical cores. 2. Set the lowest CPU and I/O priorities for emerge: PORTAGE_NICENESS=19 and run emerge as `ionice -c3 emerge ...`, you have to use CFQ scheduler to be able to use ionice. 3. Use zswap with z3fold allocator. It helps me well on my laptop with 2GB RAM. 4. If everything above doesn't help: a) reduce -j to a value where memory usage is suitable for you; b) consider using -Os in your {C,CXX,F,FC}FLAGS, since such system is certainly short of memory. 5. If even 4. doesn't help, consider using more powerful host to build binary packages for this one. Best regards, Andrew Savchenko pgpfJpZXTRw2Y.pgp Description: PGP signature
Re: [gentoo-user] Re: package.provided syntax for overlay
On Mon, 20 Feb 2017 14:07:39 +0100, Kai Krakow wrote: > > > Another option is to copy/symlink the specific package you want > > > from the bar overlay to your local overlay and do not include the > > > bar overlay in repos.conf. > > > > Sorry for being dense. Do you mean first add the overlay with > > 'layman -a bar', then symlink the particular package to my local > > overlay? How will I be updating this package in the future, if I do > > not have the 'bar' overlay settings > > in /etc/portage/repos.conf/layman.conf? > > > > I'm trying to understand the benefit of doing it as you suggest > > above ... :-/ > > You could package.mask */*::bar and then only unmask the needed bits. That does seem a cleaner way of doing it. When I first did this, that option wasn't available. -- Neil Bothwick Hell: Filling out the paperwork to get into Heaven. pgpZwx8JL4T4K.pgp Description: OpenPGP digital signature
[gentoo-user] Re: package.provided syntax for overlay
Am Sun, 19 Feb 2017 10:20:41 + schrieb Mick: > Hi All, > > Given sddm is not working for my setup, as per bug #608690, I thought > of trying entrance from the bar overlay. It wants to pull in > enlightenment, which I have already installed from the main tree and > would like to keep it as such: > > # emerge -uaDv entrance > > These are the packages that would be merged, in order: > > Calculating dependencies... done! > [ebuild U ~] x11-wm/enlightenment-:0.17/::bar > [0.20.6:0.17/0.20.6::gentoo] USE="eeze%* nls pam ukit -doc -egl% > -pm-utils% - static-libs -systemd -wayland (-spell%*)" > ENLIGHTENMENT_MODULES="appmenu backlight battery bluez4 clock > conf-applications conf-bindings conf-dialogs conf-display > conf-interaction conf-intl conf-menus conf-paths conf-performance > conf-randr conf-shelves conf-theme conf-window-manipulation > conf-window- remembers connman contact%* cpufreq everything fileman > fileman-opinfo gadman ibar ibox lokker mixer msgbus music-control > notification pager pager16%* quickaccess shot start syscon systray > tasks teamwork temperature tiling winlist wizard xkbswitch -access% > -packagkit% -wl-desktop-shell* -wl-drm* -wl- fb% -wl-x11* (-conf%*) > (-geolocation%*) (-packagekit%*) (-pager-plain%*) (- policy-mobile%*) > (-wl-text-input%*) (-wl-weekeyboard%*) (-wl-wl%*) (- xwayland%*)" 0 > KiB [ebuild N*] x11-plugins/entrance-::bar USE="consolekit > pam -grub - systemd -vkbd" 0 KiB > > > So I tried in /etc/portage/package.provided any combination of these: > > x11-wm/enlightenment-:0.17/::bar > > =x11-wm/enlightenment-:0.17 > > x11-wm/enlightenment- > > None of which can stop portage dragging in 'x11- > wm/enlightenment-:0.17/::bar'. What is the correct syntax to > block this version of enlightenment from emerging? The file needs to go to /etc/portage/profile/package.provided to have any effect. But I guess this is the wrong way for what you're going to accomplish. You should instead mask that version and then see, what pulls it in. Or you could try the world upgrade with "--exclude enlightenment" switch. -- Regards, Kai Replies to list-only preferred. pgpkf2CxgORn5.pgp Description: Digitale Signatur von OpenPGP
[gentoo-user] Re: puzzling behavior of USE flags with sys-apps/man-db
Am Mon, 20 Feb 2017 07:29:20 -0500 schrieb Harry Putnam: > Adding the X useflag to xll-libs/gtk+ > > # cat /etc/portage/package.use/gtk+ > xll-libs/gtk+ X > > But emerge appears not to go by its own stipulation Maybe because it should be "x11" and not "xll"? -- Regards, Kai Replies to list-only preferred.
[gentoo-user] Re: package.provided syntax for overlay
Am Sun, 19 Feb 2017 11:17:11 + schrieb Mick: > On Sunday 19 Feb 2017 10:50:31 Neil Bothwick wrote: > > On Sun, 19 Feb 2017 11:45:27 +0100, Johannes Rosenberger wrote: > [...] > > > > > > According to the portage manpage 'x11-wm/enlightenment-' > > > should be the correct syntax. > > > > > > But I think, package.provided is the wrong file at all. The > > > correct way to accomplish what you want to is masking > > > 'x11-wm/enlightenment-:0.17/::bar'. > > > > Agreed. > > > > Another option is to copy/symlink the specific package you want > > from the bar overlay to your local overlay and do not include the > > bar overlay in repos.conf. > > Sorry for being dense. Do you mean first add the overlay with > 'layman -a bar', then symlink the particular package to my local > overlay? How will I be updating this package in the future, if I do > not have the 'bar' overlay settings > in /etc/portage/repos.conf/layman.conf? > > I'm trying to understand the benefit of doing it as you suggest > above ... :-/ You could package.mask */*::bar and then only unmask the needed bits. -- Regards, Kai Replies to list-only preferred. pgpbWadLvkH5f.pgp Description: Digitale Signatur von OpenPGP
[gentoo-user] Re: Collision between app-arch/lrzip and net-dialup/lrzsz
Am Sun, 19 Feb 2017 15:53:46 +0100 schrieb meino.cra...@gmx.de: > Hi, > > Both packages install the files: > > * Detected file collision(s): > * > */usr/bin/lrz > */usr/share/man/man1/lrz.1.bz2 > * > > I IxQuicked for that but didn't find anything useful. > Is there a gentle way out of this collision or is > this a typical Highlander situation? Since lrzip binaries seem to be part of lrzsz, I would simply ditch that package. If it is pulled in by another ebuild statically (instead of allowing either the one or the other), it should be reported. Maybe a virtual should be provided for this. Meanwhile, you could temporary fix such a situation by putting lrzip into /etc/portage/profile/package.provided and let --depclean do the rest. -- Regards, Kai Replies to list-only preferred.
Re: [gentoo-user] puzzling behavior of USE flags with sys-apps/man-db
On Mon, 20 Feb 2017 07:29:20 -0500, Harry Putnam wrote: > Adding the X useflag to xll-libs/gtk+ > > # cat /etc/portage/package.use/gtk+ > xll-libs/gtk+ X > > But emerge appears not to go by its own stipulation > > Same output and next attempt: > > These are the packages that would be merged, in order: > > Calculating dependencies... done! > > !!! The ebuild selected to satisfy "x11-libs/gtk+:3" has unmet > requirements. > > - x11-libs/gtk+-3.22.5::gentoo USE="introspection -X (-aqua) -broadway > -cloudprint -colord -cups -examples -test -vim-syntax -wayland > -xinerama" ABI_X86="64 -32 -x32" > > The following REQUIRED_USE flag constraints are unsatisfied: any-of > ( aqua wayland X ) Either you have a type in your package.use setting or something else is unsetting X for this package. I'd start by looking at grep -r X /etc/portage -- Neil Bothwick "One world, one web, one program" -- Microsoft promotional ad "Ein Volk, ein Reich, ein Fuehrer" -- Adolf Hitler pgpyKF8UysTFf.pgp Description: OpenPGP digital signature
[gentoo-user] Re: How to dump kde gracefully in favor of lxde
Am Sat, 18 Feb 2017 22:05:01 -0500 schrieb "Walter Dnes": > On Sat, Feb 18, 2017 at 04:57:52PM -0500, Harry Putnam wrote > > > Any advice about slick ways of getting fully updated but dumping kde > > on the way. > > *IMPORTANT* KDE is obscene about dependancies. E.g. when a > lightweight pdf-reader was phased out, I looked at various options > including okular. It's an "itty-bitty-little-applet"... that seems to > pull in 90% of KDE as dependancies. If you want to get rid of KDE, > you must be prepared to dump every last little KDE app/applet. It's > an all-or-nothing situation. Sorry. > > 1) "eselect profile list" and switch to a basic non-KDE profile of > your choice. > > 2) "emerge gentoolkit" if not already present. > > 3) "cat /var/lib/portage/world" and see what KDE stuff you have. > > 4) Unmerge (i.e. "emerge --unmerge) obvious KDE-related stuff that you > find in world. > > 5) "emerge --depclean" (May not help if you've done "emerge --sync" > and not fully updated). > >The next 3 steps are going to be repeated several times > > 6) "emerge -pv --changed-use --deep --update @world" > > 7) You'll probably see portage try to pull KDE back in. For each lib > "fu-bar/foobar" that portage tries to pull in do "equery d > fu-bar/foobar" and manually unmerge whatever it finds. (Note: > gentoolkit provides the equery tool). > > 8) GOTO 6 (until portage stops trying to pull in KDE stuff). You could also try to stop portage from even syncing the KDE components into the tree. I usually do this for small systems to not give portage any chance of pulling in unwanted components. As a plus, syncing and dep calculation should be faster. https://wiki.gentoo.org/wiki/Handbook:X86/Portage/CustomTree#Excluding_packages_and_categories -- Regards, Kai Replies to list-only preferred.
[gentoo-user] puzzling behavior of USE flags with sys-apps/man-db
Setup: VBox vm running gentoo(amd64) guest on a win-10 (64bit) host Hardware: HP xw8600 - 2x Xeon CPU X5450 @ 3.00GHz - 32 GB ram On first attempt at emerging sys-apps/man-db (in came up in a world update (including -N [--newuse]) emerge's output indicated that one could not have both berkdb and gmdb so I put this in /etc/portage/package.use/man-db sys-app/man-db berkdb -gmdb That seemed to resolve that part of the problem On next attempted emerge -va sys-apps/man-db I started getting this: root # emerge -vaDN sys-apps/man-db These are the packages that would be merged, in order: Calculating dependencies... done! !!! The ebuild selected to satisfy "x11-libs/gtk+:3" has unmet requirements. - x11-libs/gtk+-3.22.5::gentoo USE="introspection -X (-aqua) -broadway -cloudprint -colord -cups -examples -test -vim-syntax -wayland -xinerama" ABI_X86="64 -32 -x32" The following REQUIRED_USE flag constraints are unsatisfied: any-of ( aqua wayland X ) The above constraints are a subset of the following complete expression: any-of ( aqua wayland X ) xinerama? ( X ) Note the part: any-of (aqua wayland X) Ok so I did echo "xll-libs/gtk+ X" > /etc/portage/package.use/gtk+ Adding the X useflag to xll-libs/gtk+ # cat /etc/portage/package.use/gtk+ xll-libs/gtk+ X But emerge appears not to go by its own stipulation Same output and next attempt: These are the packages that would be merged, in order: Calculating dependencies... done! !!! The ebuild selected to satisfy "x11-libs/gtk+:3" has unmet requirements. - x11-libs/gtk+-3.22.5::gentoo USE="introspection -X (-aqua) -broadway -cloudprint -colord -cups -examples -test -vim-syntax -wayland -xinerama" ABI_X86="64 -32 -x32" The following REQUIRED_USE flag constraints are unsatisfied: any-of ( aqua wayland X ) The above constraints are a subset of the following complete expression: any-of ( aqua wayland X ) xinerama? ( X ) No change.. if I insert `waylan' as use flag instead of X, same result and same output. I'm guessing I must be interpreting the emerge output wrong. Can anyone offer a clue here?
[gentoo-user] Re: How to keep my system from (nearly) freezing?
Am Sun, 19 Feb 2017 14:11:29 +0100 schrieb Miroslav Rovis: > On 170219-13:53+0100, Helmut Jarausch wrote: > > Hi, > > > > sometime I have some memory hungry ebuilds in the background, when > > I > Ebuilds are just text files, they don't run in the background... > > > start (e.g.) Chromium which needs very much memory if you have a > > lot of open tabs. > ( Chromium is the most privacy-invading browser ever. It's a spyware, > I could never use it, but forget about that, it's not what this topic > is about... ) > > In that case my system nearly freezes. I cannot even kill chrome. > > What can I do in that case. (Remote login doesn't work either) > Try Ctrl+Alt+Fx > where x is one of F1 ... F6 > and then issue: > # killa chromium > > > > Can I have any additional program (like Chromium) die if there is > > not enough memory. > > > > Many thanks for a hint, > > Helmut > > > This could be a hardware, not a software issue. Also, not sure, but > looks like, not a memory issue, but a CPU issue. > > It's likely the CPU triggers the BIOS to shut down because CPU gets > too hot, but because it is not properly implemented, what happens is > even worse than doing nothing, and that is: the system freezes, but > the CPU keeps running... Bad! > > How warm does you machine, try to touch it in the back, or under, if > it's a laptop, where ther CPU is? > > It reminds me of what I had. My systems, that had only the original, > run-of-the-mill coolers on the CPUs (I bought a few of same model MBO, > so i can clone my systems)... The usual 80mm coolers. > > As soon as I replaced them with 120mm coolers, no issues any more. I can second this. An optimized parallel emerge is able to keep your cores fully busy at 100% almost all of the time. All components of the cores are needed, so there's a lot of heat produced (artificial benchmarks usually cannot do the same). You should see lines about machine check exceptions (MCE) or CPU throttling logged into dmesg if this hits you. CPU throttling is a real performance killer (ebuilds suddenly take 10x the time) and may well explain what you are seeing. Getting bigger fans and coolers can do magic here, even the cheap ones. I've only lately swapped the stock cooler for a Mars T1 and could now set the turbo frequency limit back to 4.2 GHz without seeing these messages during emerge (3.7 stock limit). Before, I had to turn off turbo. I think the stock cooler was just aging (and dusted, couldn't be removed, stuck like glue to the cooler). So, check your cooler, blow the dust away, check the fan if it still easily rotates when pushed with a finger while turned off. It helps a lot. Maybe swap it with a better cooler/fan if you're experiencing throttling (I recommend it). -- Regards, Kai Replies to list-only preferred. pgpZPKVux7YqC.pgp Description: Digitale Signatur von OpenPGP
[gentoo-user] Re: How to keep my system from (nearly) freezing?
Am Sun, 19 Feb 2017 14:41:20 +0100 schrieb meino.cra...@gmx.de: > Helmut Jarausch[17-02-19 14:04]: > > Hi, > > > > sometime I have some memory hungry ebuilds in the background, when > > I start (e.g.) Chromium which needs very much memory if you have a > > lot of open tabs. > > > > In that case my system nearly freezes. I cannot even kill chrome. > > What can I do in that case. (Remote login doesn't work either) > > > > Can I have any additional program (like Chromium) die if there is > > not enough memory. > > > > Many thanks for a hint, > > Helmut > > > > Hi Helmut, > > I know that situation very well...additionally I have Blender > open... > > But I think that the "freeze" of the system is not due to the memory > amount but due to the heavy I/O while swapping. Yes, and it's the small tiny write IO every application does every now and then which then blocks the application or even the whole OS for minutes. Because IO work-queues tend to be really big. > May be a tool like ionice could help you to keep the possibility > of killing certain processes. Ionice the emerge itself and additinally > nice it also. No, ionice cannot help here. If a process blocks due to exhausting dirty_background_bytes, it is blocked. Nothing will help to write back that data any earlier. And if even dirty_bytes is exhausted, the whole system blocks. I suggested another approach in my other reply: Lower dirty_background_bytes, maybe switch to deadline scheduler for better in-time servicing of write requests. > The emerge may take longer, but a frozen system is even slower... > ;) Not really, most of it can run from cache (if you allow for "swappiness"). It will have almost no impact on system performance if running in tmpfs. I'm having 16GB of RAM and allow portage to use 32GB of tmpfs - that means: Parts of ongoing big emerges WILL be swapped out. But since using deadline, this has almost no impact. Emerging is still lots faster than without tmpfs. -- Regards, Kai Replies to list-only preferred.
[gentoo-user] Re: How to keep my system from (nearly) freezing?
Am Sun, 19 Feb 2017 13:53:49 +0100 schrieb Helmut Jarausch: > Hi, > > sometime I have some memory hungry ebuilds in the background, when I > start (e.g.) Chromium which needs very much memory if you have a lot > of open tabs. > > In that case my system nearly freezes. I cannot even kill chrome. > What can I do in that case. (Remote login doesn't work either) > > Can I have any additional program (like Chromium) die if there is > not enough memory. You may want to switch to the deadline scheduler and see if it improves things. For my setup, it did (bcache involved). But I'm pretty sure my usage pattern is a bit different, I'm emerging in a tmpfs (I have 16 GB of RAM), so every big ebuild probably forces data going to swap. Actually, all my three hard disks have a 10 GB swap partition in front with same priority. The file system itself is cached through bcache, so it should be mostly decoupled (this means, put swap to mostly idle disks). But I was still seeing extensive freezes until I switched to the deadline scheduler. Kernel 4.10, which brings write-back throttling, will probably help also (and maybe I can switch back to CFQ then to benefit from ioprio again). The problem is something like buffer bloat known from networking: Writes queue up in a long queue, and even small writes will block your applications. CFQ tends to build really long such queues while it tries to maintain "fairness". Deadline instead tries to service write requests as fast as possible after a short timeout (which can be configured in sysfs). Even then, a huge queue can still pile up, so I suggest you reduce dirty_background_bytes to a sane value (the default percent values really don't play well with amounts of RAM installed today): vm.dirty_background_bytes = 134217728 This makes processes with dirty data block much earlier and should reduce the queue. You may want to lower it even more and see how your system behaves. There's a chance that these shorter queues can be written before the whole system blocks (because dirty_bytes is exhausted). At least it should reduce the time until the system becomes responsive again. I think that swap space is there and is cheap to use most of the time, so you should encourage your system to use it and use it early (prefer caching over parking dead memory blocks in RAM). The problem comes when your system is already stalled by other IO - which apparently is the case most of the time when the system needs to start using swap. The only fix is to keep the IO queues short. The knobs mentioned above should help you with that. -- Regards, Kai Replies to list-only preferred.
Re: [gentoo-user] Bluefish colours
On Sunday 19 Feb 2017 16:58:19 Mick wrote: > I don't have an answer for the bluefish button colours, but just an idea > that different Gnome themes may be also applied on the Bluefish > application since it uses Gtk. You've done it again, Mick. Bluefish didn't like clearlooks-phenix, but it's quite happy with breeze. > Regarding your Chromium problem, have you configured your > kernel/firmware/x11 video drivers to use hardware acceleration? Chromium > uses hardware acceleration if available and the problem of blank pages > may be relevant, but I am not sure. Try this in the address bar to see > what it reports: > > chrome://gpu After a reboot on a freshly compiled kernel my Chromium problem has gone away. Thanks for your help. -- Regards Peter
Re: [gentoo-user] emerge a binary package no longer in tree
Neil Bothwick wrote: On Mon, 20 Feb 2017 07:50:15 +0100, Raffaele Belardi wrote: I'd like to try and update a package I masked long time ago due to performance problems. Upstream the problem does not seem completely addressed and solved so I'd like to be able to go back to the old version just in case. But the old version is no longer in the tree. If I quickpkg it will I be able to reinstall it anyway? Yes, by passing the path to the package to emerge. But this is considered experimental, a safer solution is to copy the ebuild to a local overlay. The ebuilds for all installed packages are in /var/db/pkg. Great hint about the ebuild copy in /var/db/pkg, thanks. raffaele
Re: [gentoo-user] emerge a binary package no longer in tree
On Mon, 20 Feb 2017 07:50:15 +0100, Raffaele Belardi wrote: > I'd like to try and update a package I masked long time ago due to > performance problems. Upstream the problem does not seem completely > addressed and solved so I'd like to be able to go back to the old > version just in case. But the old version is no longer in the tree. If > I quickpkg it will I be able to reinstall it anyway? Yes, by passing the path to the package to emerge. But this is considered experimental, a safer solution is to copy the ebuild to a local overlay. The ebuilds for all installed packages are in /var/db/pkg. -- Neil Bothwick I stayed up all night playing poker with tarot cards. I got a full house and four people died. pgpvNGMaAM19P.pgp Description: OpenPGP digital signature