Re: [gentoo-user] Grub, gpt partitions and BIOS, not uefi thing.

[Mickaël Bucas]

Hi

Le sam. 27 avr. 2024 à 18:53, Dale  a écrit :
> Howdy,
>
> I'm installing Gentoo on another old box.
I was wondering how old this box could be and if it had a BIOS with
UEFI and GPT.

I didn't find a precise date for BIOS, but Wikipedia[1] shows that the
first version of Windows for x64 that can read and write GPT was
published on 2005-04-25. To boot with UEFI, a later version was
published on 2006-07-22.
I think this means most BIOSes were compatible to various degrees at this time.

So if your box is less than 20 years old, it should be OK !
I don't remember how powerful the boxes were at this time, but they
still had floppy disk drives :)

> Thanks to anyone who has a link, some notes or something.  :-D
>
> Dale
>
> :-)  :-)

Good luck

Mickaël Bucas

[1] https://en.wikipedia.org/wiki/GUID_Partition_Table#Windows:_64-bit_versions

Re: [gentoo-user] Terminal emulator to replace Konsole

[Mickaël Bucas]

Le ven. 22 mars 2024 à 21:02, Dale  a écrit :
>
> Howdy,
>
> I've been using Konsole, part of KDE, for command line stuff ever since
> I started using Linux.  Linux is all I've ever used.  No windoze.  ;-)
> While Konsole is good enough for almost everything, there is one feature
> I wish it had.  The ability to edit with the mouse.  I don't know of a
> way to make it do this.  The only way I know of to edit a command, left
> arrow to what you want to edit and change it.  I'd like to find one
> where I can use the mouse to place the cursor and edit from there.  Even
> maybe highlight and replace.  As far as I know, Konsole doesn't have
> that ability.
>
> I looked in x11-terms and there is a few options, I think.  I tried
> looking at home pages and such but none of them mention a feature like
> this but it may have it.  I was wondering if anyone knows of a terminal
> emulator that allows the mouse to place the cursor to edit parts or
> whole sections of a command.  Some of my commands are really long and it
> seems the part I want to edit is always at the beginning.  :/
>
> Hoping for some ideas.
>
> Thanks.
>
> Dale
>
> :-)  :-)
>
Hi

I think it's not a terminal emulator feature, but rather a shell feature.

Some terminal programs are designed to interact with the mouse, but
bash command line, based on readline, doesn't react to mouse clicks.

I've tried Midnight Commander in Konsole and xterm, and it actually
moves the cursor to the click position in its own command line !

Maybe there's an extension or set of parameters for bash or other
shells to handle mouse clicks, but I'm not aware of it.

Best regards
Mickaël Bucas

Re: [gentoo-user] is a global use flag necessary for python?

[Mickaël Bucas]

0::gentoo, installed) USE="userland_GNU"
> ABI_X86="(64)" PYTHON_SINGLE_TARGET="python3_8 -python3_9"
> 
>  
> >=dev-lang/python-exec-2:2/2=[python_targets_python3_8(-),python_targets_python3_9(-)]
>  required by (dev-python/pytz-2021.1:0/0::gentoo, installed) 
> USE="userland_GNU" ABI_X86="(64)" PYTHON_TARGETS="python3_8 python3_9 
> (-pypy3) -python3_10"
>
>
>
> It may be possible to solve this problem by using package.mask to
> prevent one of those packages from being selected. However, it is also
> possible that conflicting dependencies exist such that they are
> impossible to satisfy simultaneously.  If such a conflict exists in
> the dependencies of two different packages, then those packages can
> not be installed simultaneously. You may want to try a larger value of
> the --backtrack option, such as --backtrack=30, in order to see if
> that will solve this conflict automatically.
>
> For more information, see MASKED PACKAGES section in the emerge man
> page or refer to the Gentoo Handbook.
>
>
> !!! The following installed packages are masked:
> - sys-libs/glibc-2.33-r1::gentoo (masked by: package.mask)
> /var/db/repos/gentoo/profiles/package.mask:
> # Andreas K. Hüttel  (2017-05-21)
> # (and others, updated later)
> # These old versions of toolchain packages (binutils, gcc, glibc) are no
> # longer officially supported and are not suitable for general use. Using
> # these packages can result in build failures (and possible breakage) for
> # many packages, and may leave your system vulnerable to known security
> # exploits.
> # If you still use one of these old toolchain packages, please upgrade (and
> # switch the compiler / the binutils) ASAP. If you need them for a specific
> # (isolated) use case, feel free to unmask them on your system.
>
> - sys-kernel/linux-firmware-20201022-r2::gentoo (masked by: || ( )
> linux-fw-redistributable no-source-code license(s))
> A copy of the 'linux-fw-redistributable' license is located at
> '/var/db/repos/gentoo/licenses/linux-fw-redistributable'.
>
> A copy of the 'no-source-code' license is located at
> '/var/db/repos/gentoo/licenses/no-source-code'.
>
> - dev-python/docutils-0.17.1::gentoo (masked by: package.mask)
> - dev-build/cmake-3.22.2::gentoo (masked by: CMake license(s))
> - sys-devel/binutils-2.37_p1-r2::gentoo (masked by: package.mask)
> - virtual/libcrypt-1-r1::gentoo (masked by: package.mask)
> /var/db/repos/gentoo/profiles/base/package.mask:
> # Sam James  (2021-11-22)
> # Mask the older libcrypt virtual (which accepted glibc[crypt]) to ease
> # dependency resolution. In a fair number of cases, this has helped
> # upgrades go through cleanly.
> # Read the news item if you need help!
> # (This mask is undone in musl profiles where the transition is not yet
> being
> # made.)
> # bug #699422.
>
> - sys-libs/binutils-libs-2.37_p1-r2::gentoo (masked by: package.mask)
> - dev-libs/openssl-1.1.1n::gentoo (masked by: package.mask)
> /var/db/repos/gentoo/profiles/package.mask:
> # Sam James  (2023-09-09)
> # OpenSSL 1.1.x is EOL on 2023-09-11. Please upgrade immediately to >=
> OpenSSL 3.
> # https://www.openssl.org/blog/blog/2023/03/28/1.1.1-EOL/
> # https://www.openssl.org/blog/blog/2023/06/15/1.1.1-EOL-Reminder/
> # Please run a full world upgrade, especially checking /etc/portage and
> your world file
> # for old PHP or Ruby references.
>
> - media-fonts/font-bh-ttf-1.0.3-r2::gentoo (masked by: bh-luxi license(s))
> A copy of the 'bh-luxi' license is located at
> '/var/db/repos/gentoo/licenses/bh-luxi'.
>
> - media-fonts/font-bh-type1-1.0.3-r2::gentoo (masked by: bh-luxi license(s))
> For more information, see the MASKED PACKAGES section in the emerge
> man page or refer to the Gentoo Handbook.
>
>
>

It seems you have a lot of updates to handle before Firefox.

The first hint is the mask on glibc-2.33-r1, back from 2017. This is a
central package and must be updated early.
The current stable version is sys-libs/glibc-2.38-r10
As most packages are linked to glibc, they will also probably need an update.

The second hint is the mask on openssl-1.1.1n. This should also be
updated early.
The current stable version is dev-libs/openssl-3.0.13.
As above, many packages, including Firefox, depend on it.

The most simple way would be to recompile everything with "emerge -e
@world", but that's also really time consuming, and conflicts are not
solved magically...
Maybe a first step with "emerge -e @system" could take care of the
most important packages first.
You can try these commands with "--pretend --backtrack=20" to see if
you can go forward or if other blockers remain.

Best regards
Mickaël Bucas

Re: [gentoo-user] updating remote system

[Mickaël Bucas]

Le mer. 12 avr. 2023 à 11:27, Peter Humphrey  a écrit :
>
> On Wednesday, 12 April 2023 08:16:05 BST Mickaël Bucas wrote:
>
> > I don't know if it's correct, I'd proceed this way:
> > 1. Update world first, because some updated packages may be involved
> > in kernel building
> > 2. Reboot with the current working kernel and check everything else works
> > 3. Update the kernel and reboot (keeping the old kernel for a while)
>
> Perhaps you missed the point that some of the world update requires the
> updated kernel.
Thanks for pointing this out, I missed what Thelma stated clearly.

> I'd do the same as Thelma.
With VBox needs, I agree
There may be more elaborate moves like Michael explains, though I'm
not able to give more advice.

>
> --
> Regards,
> Peter.

Best regards
Mickaël

Re: [gentoo-user] updating remote system

[Mickaël Bucas]

Le mer. 12 avr. 2023, 02:26, jul...@jroy.ca  a écrit :
>
> On Tue, 2023-04-11 at 20:24 -0400, jul...@jroy.ca wrote:
> > On Tue, 2023-04-11 at 18:21 -0600, the...@sys-concept.com wrote:
> > > What is the correct way of updating remote system?
> > > I'll need to update the kernel + world
> > >
> > > 1.) Update all packages in world first, reboot and update kernel
> > > 2.) Update kernel first, reboot and update world
> > >
> > > It seems to me logical choice is 2.) as I'll need new kernel to
> > > update to new VirtualBox.
> > > Right now remote system is running linux-5.10.103-gentoo and I
> > > think
> > > new VirtualBox need kerenl-5.13 or higher.
> > >
> >
> >
> >
> > --
>
> It seems my message got sent empty...
> Let's try again:
>
> I usually update world first, then the kernel, and finally reboot.
>
>
> --
> Julien

Hi

I don't know if it's correct, I'd proceed this way:
1. Update world first, because some updated packages may be involved
in kernel building
2. Reboot with the current working kernel and check everything else works
3. Update the kernel and reboot (keeping the old kernel for a while)

That's two reboots, but for a remote system I'd keep on the safe side.

Have a nice compile time !

Mickaël

Re: [gentoo-user] Tailing compressed build logs

[Mickaël Bucas]

Le mar. 7 mars 2023 à 05:36, Bryan Gardiner  a écrit :
>
> Hi folks,
>
> How can I follow Portage's compressed build logs in real time as they
> are generated?
>
> I keep build logs and use FEATURES=compress-build-logs so that they
> don't get too large.  I can peek at how a build is going with zless on
> build.log.gz, which doesn't update (understandably), but I would
> really like to be able to watch a log with some "tail -f" equivalent.
> I get streaming output with
>
> tail -c +1 -f build.log.gz | od -t x1
>
> but the following hangs with no output:
>
> tail -c +1 -f build.log.gz | gunzip
>
> even with a build log that is 72KB compressed (2.4MB uncompressed),
> which should be larger than any pipe buffers...  Any idea why gunzip
> can't handle this, or what I should I should be doing instead?
>
> Thanks,
> Bryan
>
Hi

Reading the man page, "zless" is just a wrapper around "less".
You can check with:
$ file $(which zless)
/usr/bin/zless: POSIX shell script, ASCII text executable
$ less $(which zless)

So it should support the same options including typing "F" at the end
of a file to keep trying to read when the end of file is reached.

I made a small test, but it didn't work:
# Create a growing file
$ yes | nl | gzip > zless-test.gz &
# Try to follow at the end
$ zless zless-test.gz

With ">" to go to the end and "F" to continue, I didn't get the
expected behavior, it stood still at the point I was viewing.
I don't know if it's really a bug or if I made a mistake...
(Don't forget to stop the growing file :) )

Best regards

Mickaël Bucas

Re: [gentoo-user] Any way to automate login to host and su to root?

[Mickaël Bucas]

Le jeu. 14 juil. 2022 à 08:35, J. Roeleveld  a écrit :
>
> Hi All,
>
> I am looking for a way to login to a host and automatically change to root
> using a password provided by an external program.
>
> The root passwords are stored in a vault and I can get passwords out using a
> script after authenticating.
>
> Currently, I need to do a lot of the steps manually:
> ssh @
> su -
> (copy/paste password from vault)

Why not use directly ssh root@ ?
With an SSH key protected by a passphrase that would be a single step
to connect.
You would have a passphrase to manage but you already are using a tool for that.
If you accept the risks, you could also use an SSH key without a passphrase.

sshd on the host must be configured with
PermitRootLogin=prohibit-password at minimum, which is the default
value.

> I would like to change this to:
>  
>
> Does anyone have any hints on how to achieve this without adding a "NOPASSWD"
> entry into /etc/sudoers ?
>
> Thanks in advance,
>
> Joost

Best regards

Mickaël Bucas

Re: [gentoo-user] Does gcc 11.3 need a special/new kernel version?

[Mickaël Bucas]

Le dim. 5 juin 2022 à 09:49, Matthias Hanft  a écrit :

> Hi,
>
> I have a rather old server which I still keep alive (u never
> know if it's needed again), so I update all packages about
> once a month with
>
> emerge -aNDuv --keep-going --backtrack=999 --with-bdeps=y @world
>
> which works fine - until now. I can't upgrade gcc from 11.2.1
> to 11.3.0 because of
>
> make[3]: Entering directory
> '/var/tmp/portage/sys-devel/gcc-11.3.0/work/build/gcc'
> build/genautomata
> /var/tmp/portage/sys-devel/gcc-11.3.0/work/gcc-11.3.0/gcc/common.md
> /var/tmp/portage/sys-devel/gcc-11.3.0/work/gcc-11.3.0/gcc/config/i386/i386.md
> \
>   insn-conditions.md > tmp-automata.c
> make[3]: *** [Makefile:2456: s-automata] Error 139
> make[3]: Leaving directory
> '/var/tmp/portage/sys-devel/gcc-11.3.0/work/build/gcc'
> make[3]: *** Waiting for unfinished jobs
>
> and dmesg says
>
> [ 1297.247619] genautomata[3301]: segfault at bfea1ffc ip b76bb23b sp
> bfea2000 error 6 in genautomata[b76b1000+44000]
>
> Granted, it is still kernel 4.0.5, but I don't want to go through
> the trouble of installing a new kernel on the old system. But the
> old kernel is the only thing I can think of that could be causing
> the problem...?!
>
> Possibly there was a discussion about this, in the year 2008:
> https://gcc.gnu.org/pipermail/gcc-patches/2008-November/251389.html
> but I don't understand what's it all about, and how could I modify
> emerge to work around the problem.
>
> gcc is the only package which won't upgrade - everything else works
> fine (including glibc 2.34-r13 and all the other stuff).
>
> Any hints?
>
> Thanks,
>
> -Matt
>
>
> Hi Matthias

It seems that the mail you found resulted in the creation of a bug in GCC
Bugzilla, with the author of the mail also commenting on the bug.
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=38052

This bug is in status RESOLVED FIXED, so either it wasn't really fixed and
your situation triggers it, or it's a different issue.

It seems that the command line to compile "genautomata" contains useful
information to understand the cause.
Can you find it in the GCC compile logs ?

With that you may open a bug in Gentoo or GCC Bugzilla.

Best regards

Mickaël Bucas

Re: [gentoo-user] strange ebuild problem

[Mickaël Bucas]

Le ven. 20 nov. 2020 à 16:44, Helmut Jarausch  a écrit :

> Hi,
> I have an ebuild containing
>
> inherit git-r3
>
> and
>
> src_compile() {
>
> ./build_all.sh
> }
>
> The script build_all.sh contains
>
> [ -d csources ] || git clone --depth 1
> https://github.com/nim-lang/csources.git
>
>
> When I try to build I get the strange error
> fatal: unable to access 'https://github.com/nim-lang/csources.git/':
> Could not resolve host: github.com
>
> But when I exec
> git clone --depth 1 https://github.com/nim-lang/csources.git
> in an xterm (i.e. without using 'ebuild') it succeeds.
>
> What's going on here?
>
> Many thanks for a hint,
> Helmut
>

Hi Helmut

I believe this is a consequence of the sandboxing of ebuilds.

When I had the same problem with an ebuild trying to download many things
from the Maven repository, I added the following lines inside the ebuild:
# To enable Maven access to https://repo.maven.apache.org/maven2
RESTRICT="network-sandbox"

It is described (shortly) in "man 5 ebuild"
This is forbidden in the Portage tree, but is allowed in your own overlay.

Best regards
Mickaël Bucas

Re: [gentoo-user] soft keyboard for touchscreen

[Mickaël Bucas]

Le dim. 18 oct. 2020 à 19:17, William Kenworthy  a
écrit :

> Thanks Mickaël,
>
> I have onboard installed from an overlay (I have tried a couple of
> different ones, they seem to be variants of the same original package)
>
> How are you starting onboard? - the packages do not come with an
> initscript or hooks into X that I can see (though they seem to depend on
> systemd - I use openrc)
>
> BillK
>
I start it manually, I've never thought about automating it :)
On some systems I'm sometimes physically connected with a real keyboard and
I don't need it in this situation.

Best regards
Mickaël Bucas

> On 19/10/20 12:46 am, Mickaël Bucas wrote:
>
> Hi William
>
> Le dim. 18 oct. 2020 à 03:05, William Kenworthy  a
> écrit :
>
>> Can someone recommend a guide to installing a touch screen aware soft
>> keyboard in gentoo?
>>
>> I have tried a number of keyboards but the various guides do not say how
>> to integrate a soft keyboard in to a window manager (I am using xfwm4
>> but could change) or login screen.
>>
>> I can manually start them, but they do not show up when an editor,  text
>> box or login is required so I have to attach a physical keyboard to
>> regain control.
>>
>>
>> BillK
>>
>
>
> If you're using SDDM, you can activate the virtual keyboard with the
> following line in the config file /etc/sddm.conf
>
> [General]
> # Input method module
> InputMethod=qtvirtualkeyboard
>
> For the session itself, under KDE on Ubuntu I use Onboard [1], which is
> not available in Portage
> I found a blog page [2] explaining how to install Onboard on Gentoo, with
> an ebuild, but it's for Python 3.4 to 3.6
> I've adapted the ebuild to Python 3.7 in my overlay [3] and it worked as
> expected, either from another PC through VNC or from a touch screen.
> Onboard seems to originate from Gnome, and as I use it under KDE, it
> should be independent from the window manager.
>
> Best regards
> Mickaël Bucas
>
> [1] https://launchpad.net/onboard
> [2]
> https://fitzcarraldoblog.wordpress.com/2018/06/25/installing-the-onboard-on-screen-keyboard-in-gentoo-linux/
> [3] https://github.com/mbucas/gentoo-overlay
>
>

Re: [gentoo-user] soft keyboard for touchscreen

[Mickaël Bucas]

Hi William

Le dim. 18 oct. 2020 à 03:05, William Kenworthy  a
écrit :

> Can someone recommend a guide to installing a touch screen aware soft
> keyboard in gentoo?
>
> I have tried a number of keyboards but the various guides do not say how
> to integrate a soft keyboard in to a window manager (I am using xfwm4
> but could change) or login screen.
>
> I can manually start them, but they do not show up when an editor,  text
> box or login is required so I have to attach a physical keyboard to
> regain control.
>
>
> BillK
>


If you're using SDDM, you can activate the virtual keyboard with the
following line in the config file /etc/sddm.conf

[General]
# Input method module
InputMethod=qtvirtualkeyboard

For the session itself, under KDE on Ubuntu I use Onboard [1], which is not
available in Portage
I found a blog page [2] explaining how to install Onboard on Gentoo, with
an ebuild, but it's for Python 3.4 to 3.6
I've adapted the ebuild to Python 3.7 in my overlay [3] and it worked as
expected, either from another PC through VNC or from a touch screen.
Onboard seems to originate from Gnome, and as I use it under KDE, it should
be independent from the window manager.

Best regards
Mickaël Bucas

[1] https://launchpad.net/onboard
[2]
https://fitzcarraldoblog.wordpress.com/2018/06/25/installing-the-onboard-on-screen-keyboard-in-gentoo-linux/
[3] https://github.com/mbucas/gentoo-overlay

Re: [gentoo-user] Portage: Complaining about multiple incidents at once

[Mickaël Bucas]

Le dim. 19 avr. 2020 à 13:03, Gerion Entrup  a écrit :
>
> Hi,
>
> does Portage have a "don't stop" mode?
>
> With that I mean the following:
> I'm doing updates with
> emerge -auND world --keep-going --quiet-build --verbose-conflicts
>
> Portage then often stops, saying a dependency keyword is missing or a
> useflag etc. However, this always happens incident after incident, so I
> need to run portage multiple times.
>
> Is there a way to instruct portage to just say: "Ok, here is a
> configuration fix necessary, maybe a use flag must be added, but let's
> assume, the user does this and the use flag is set now. What do I need
> to do next."
>
> Without checking it, this behaviour seems to be more the case in prior
> portage version (complaining about multiple incidents at once).
>
> Best,
> Gerion

Hi Gerion

I believe that what you are looking for are the options starting with
--autounmask, which affect the initial analysis of the build tree,
which is the source of the complaints about missing keywords or USE
flags.
The --keep-going option does a great job at continuing to build other
packages when one build fails.

There are a bunch of them so I don't know how much you'd like to use
but it seems tailored to go forward in almost any situation !
They are well described in "man emerge"
--autounmask [ y | n ]
--autounmask-backtrack < y | n >
--autounmask-continue [ y | n ]
--autounmask-only [ y | n ]
--autounmask-unrestricted-atoms [ y | n ]
--autounmask-keep-keywords [ y | n ]
--autounmask-keep-masks [ y | n ]
--autounmask-license < y | n >
--autounmask-use < y | n >
--autounmask-write [ y | n ]

Adding --backtrack=COUNT with COUNT > 10 (the default value) may push
Portage to look forward a bit more.

However, as I don't use them, I don't know what stability you could
expect from your system after activating any or all of them.

Best regards
Mickaël Bucas

Re: [gentoo-user] remote console

[Mickaël Bucas]

Hi

I'm using Shell In A Box, which is available in Portage : www-misc/shellinabox
https://github.com/shellinabox/shellinabox

As it uses /bin/login, you can assign users to your students so they
are independent of each other.

Best regards

Mickaël Bucas

Le mar. 24 mars 2020 à 07:51, J. Roeleveld  a écrit :
>
> On Tuesday, March 24, 2020 4:54:47 AM CET William Kenworthy wrote:
> > Hi,
> >
> >  I am giving a PD session to some techs on raspberry pi's and odroid
> > arm systems running gentoo.  It will be online through MS teams which
> > works well through the google chrome browser using WebRTC.  I would like
> > to use browser based ssh to look at various files, logs etc. using vi.
> > Can someone recommend some type of browser based ssh client that will
> > run in chrome?
> >
> > MS Teams has a linux app, but its less functional than the browser
> > version - I cant get the camera or desktop export to work.
> >
> > BillK
>
> something like this?
>
>
> http://web-console.org/
> (Not sure, but might be providing access as the webserver-user)
>
> or:
> https://pypi.org/project/webssh/
> (This looks more promising. The screenshot also mentions Raspbary Pi)
>
> I have not tried either of these, please let me know what they are like as I
> am thinking of adding this myself to some services.
>
> Many thanks,
>
> Joost
>
>
>

Re: [gentoo-user] Migrating chroot to VM

[Mickaël Bucas]

Le jeu. 5 mars 2020 à 13:07, Michael  a écrit :
>
> On Thursday, 5 March 2020 09:54:01 GMT Alessandro Barbieri wrote:
> > I want to do a crazy thing. I want to migrate my gentoo installation from a
> > chroot to a VM (both on the same host). The host is debian but I can only
> > use SSH. What do you suggest to do?
>
> There are VMWare applications to convert a physical installation of an OS to a
> VM disk, but since the guest OS is linux, the way I would do it would be:
>
> 1. Create a virtual disk of the desired size.
> 2. Attach a LiveCD to the VM and boot with it.
> 3. Use the LiveCD environment to partition/format/mount the VM disk.
> 4. Use rsync to copy the filesystem contents from the chroot to the VM
> partitions.
> 5. Edit the copied fstab to correspond to the VM disk partition UUIDs and
> potentially reinstall GRUB.
> 6. Unmount the VM disk partitions, shutdown the VM, detatch the LiveCD and
> restart the VM.
>
> You could use partclone instead of rsync, which would retain the same UUIDs
> and filesystems as the chrooted system and therefore simplify the migration.
>
> HTH.

As a chroot is not a complete system, you will also have to do some
more actions inside the VM
- Compile and install a kernel [1], probably with the VMware modules
to get better performance.
- Tell Grub to include that kernel in the list of boot options [2]
- Check that an init system is fully configured, you can choose
between OpenRC or SystemD [3]
- Configure networking inside the VM [4] and in VMware on the host
- Add services like SSH to system start to be able to connect to the VM

As you can see, this will be a mix of a fresh install and a migration
of your chroot. Basically you can read through the Gentoo Handbook to
see what steps are necessary for the VM to boot.
When you copy the chroot files into the VM they will overwrite files
from the Stage3.
I believe the best step to do this is just before the "Installing
tools" chapter, but their may be side effects I don't anticipate...

The good thing about a VM is that if anything goes wrong, you just
trash it and restart !
Tell us how things turn out for you.

Best regards
Mickaël Bucas

[1] https://wiki.gentoo.org/wiki/Handbook:AMD64/Installation/Kernel
[2] https://wiki.gentoo.org/wiki/Handbook:AMD64/Installation/Bootloader
[3] https://wiki.gentoo.org/wiki/Handbook:AMD64/Installation/Base
[4] https://wiki.gentoo.org/wiki/Handbook:AMD64/Installation/System

Re: [gentoo-user] Stable Python package changes USE flags with ~amd64

[Mickaël Bucas]

Hi Franz

Thanks for your insight into Portage inner workings. I'm glad I learned
something !
And that bug from 6 years ago is a sign that something isn't clear about
this subject.

Thanks
Best regards
Mickaël Bucas


Le mar. 7 janv. 2020 à 17:56, Franz Fellner  a
écrit :

> OK, seems I can reproduce (had an issue with my config in a previous
> attempt).
> Probably related:
> https://bugs.gentoo.org/491166
> But your view on the matter isn't correct.
> Portage is strict when it comes to dependencies. Just because py3_7 is
> installed it won't enable the PYTHON_TARGET because you might uninstall
> python-3.7 and end up with a broken olefile.
> What actually seems to happen: python3_7 (together with other)
> PYTHON_TARGETS is disabled in the profile via use.stable.mask.
> That config file disables certain USE-Flags for stable packages. That way
> py3_7 is available for testing versions but not for stable ones.
> olefile-0.46 is only available as stable version. But now adding it to
> package.accept_keywords automagically seems to enable those
> use.stable.mask'ed USE-Flags.
> IMO this is a bug as it introduces totally unpredictable (and AFAICS
> undocumented) behaviour.
>
> Let's see what the DEVs say about this!
>
> Regards
> Franz
>
> Am Di., 7. Jan. 2020 um 18:27 Uhr schrieb Mickaël Bucas  >:
>
>> I get the following result:
>> # emerge -pv1 olefile
>>
>>
>> These are the packages that would be merged, in order:
>> Calculating dependencies... done!
>> [ebuild   R] dev-python/olefile-0.46::gentoo  USE="-doc"
>> PYTHON_TARGETS="python2_7 python3_6 (-pypy3) (-python3_7*) (-python3_8)" 0
>> KiB
>> Total: 1 package (1 reinstall), Size of downloads: 0 KiB
>>
>> It seems to be in line with the interpretation I've come up with.
>>
>> Best regards
>> Mickaël Bucas
>>
>> Le mar. 7 janv. 2020 à 16:18, Franz Fellner  a
>> écrit :
>>
>>> And what if you change the line to "dev-python/olefile amd64"?
>>>
>>> Am Di., 7. Jan. 2020 um 17:10 Uhr schrieb Mickaël Bucas <
>>> mbu...@gmail.com>:
>>>
>>>> Hi Franz
>>>>
>>>> Thanks for your reply.
>>>>
>>>> However your assumption is incorrect: these two commands are run on the
>>>> same machine, with only the keyword on "olefile" changed.
>>>> Thinking a bit more about it, Python 3.7 isn't stable yet, so I also
>>>> have "=dev-lang/python-3.7* ~amd64" in package.accept_keyword.
>>>>
>>>> I've been able to reproduce this behavior in a chroot based on stage 3
>>>> with the minimum packages installed.
>>>> I have in make.conf
>>>> PYTHON_TARGETS="python2_7 python3_6 python3_7"
>>>> In /var/lib/portage/world
>>>> dev-lang/python:3.7
>>>> dev-python/olefile
>>>> In /etc/portage/package.accept_keywords
>>>> dev-python/olefile ~amd64
>>>> =dev-lang/python-3.7* ~amd64
>>>> dev-python/setuptools ~amd64
>>>> dev-python/certifi ~amd64
>>>>
>>>> And emerge says :
>>>> # emerge -pv1 olefile
>>>> These are the packages that would be merged, in order:
>>>> Calculating dependencies... done!
>>>> [ebuild   R] dev-python/olefile-0.46::gentoo  USE="-doc"
>>>> PYTHON_TARGETS="python2_7 python3_6 python3_7 -pypy3 -python3_8" 0 KiB
>>>> Total: 1 package (1 reinstall), Size of downloads: 0 KiB
>>>>
>>>> When I remove " dev-python/olefile ~amd64", Python 3.7 would be
>>>> disabled :
>>>> # emerge -pv1 olefile
>>>> These are the packages that would be merged, in order:
>>>> Calculating dependencies... done!
>>>> [ebuild   R] dev-python/olefile-0.46::gentoo  USE="-doc"
>>>> PYTHON_TARGETS="python2_7 python3_6 (-pypy3) (-python3_7*) (-python3_8)" 0
>>>> KiB
>>>> Total: 1 package (1 reinstall), Size of downloads: 0 KiB
>>>>
>>>> This is still puzzling me, but one interpretation may be :
>>>> I you enable the unstable ~amd64 keyword on a package, the stable
>>>> version of said package is allowed to run on the unstable version of the
>>>> Python interpreter.
>>>>
>>>> This seems to be the intended behavior, as I found that at least 40
>>>> Python packages on each of my 2 systems are stable and have Python 3.7
>>>> enabled (I keyworded all of them sometime in the past...)
>>>

Re: [gentoo-user] Stable Python package changes USE flags with ~amd64

[Mickaël Bucas]

I get the following result:
# emerge -pv1 olefile


These are the packages that would be merged, in order:
Calculating dependencies... done!
[ebuild   R] dev-python/olefile-0.46::gentoo  USE="-doc"
PYTHON_TARGETS="python2_7 python3_6 (-pypy3) (-python3_7*) (-python3_8)" 0
KiB
Total: 1 package (1 reinstall), Size of downloads: 0 KiB

It seems to be in line with the interpretation I've come up with.

Best regards
Mickaël Bucas

Le mar. 7 janv. 2020 à 16:18, Franz Fellner  a
écrit :

> And what if you change the line to "dev-python/olefile amd64"?
>
> Am Di., 7. Jan. 2020 um 17:10 Uhr schrieb Mickaël Bucas  >:
>
>> Hi Franz
>>
>> Thanks for your reply.
>>
>> However your assumption is incorrect: these two commands are run on the
>> same machine, with only the keyword on "olefile" changed.
>> Thinking a bit more about it, Python 3.7 isn't stable yet, so I also have
>> "=dev-lang/python-3.7* ~amd64" in package.accept_keyword.
>>
>> I've been able to reproduce this behavior in a chroot based on stage 3
>> with the minimum packages installed.
>> I have in make.conf
>> PYTHON_TARGETS="python2_7 python3_6 python3_7"
>> In /var/lib/portage/world
>> dev-lang/python:3.7
>> dev-python/olefile
>> In /etc/portage/package.accept_keywords
>> dev-python/olefile ~amd64
>> =dev-lang/python-3.7* ~amd64
>> dev-python/setuptools ~amd64
>> dev-python/certifi ~amd64
>>
>> And emerge says :
>> # emerge -pv1 olefile
>> These are the packages that would be merged, in order:
>> Calculating dependencies... done!
>> [ebuild   R] dev-python/olefile-0.46::gentoo  USE="-doc"
>> PYTHON_TARGETS="python2_7 python3_6 python3_7 -pypy3 -python3_8" 0 KiB
>> Total: 1 package (1 reinstall), Size of downloads: 0 KiB
>>
>> When I remove " dev-python/olefile ~amd64", Python 3.7 would be disabled :
>> # emerge -pv1 olefile
>> These are the packages that would be merged, in order:
>> Calculating dependencies... done!
>> [ebuild   R] dev-python/olefile-0.46::gentoo  USE="-doc"
>> PYTHON_TARGETS="python2_7 python3_6 (-pypy3) (-python3_7*) (-python3_8)" 0
>> KiB
>> Total: 1 package (1 reinstall), Size of downloads: 0 KiB
>>
>> This is still puzzling me, but one interpretation may be :
>> I you enable the unstable ~amd64 keyword on a package, the stable version
>> of said package is allowed to run on the unstable version of the Python
>> interpreter.
>>
>> This seems to be the intended behavior, as I found that at least 40
>> Python packages on each of my 2 systems are stable and have Python 3.7
>> enabled (I keyworded all of them sometime in the past...)
>>
>> Thanks
>> Best regards
>> Mickaël Bucas
>>
>> Le mar. 7 janv. 2020 à 08:08, Franz Fellner  a
>> écrit :
>>
>>> I assume those emerge commands weren't done on one machine but come from
>>> those two different machines.
>>> This change in USE Flags can't come from that line in
>>> package.accept_keywords.
>>> This is a change in PYTHON_TARGETS in make.conf, package.use or
>>> package.env.
>>> Carefully go through those config files/directories, I am sure you will
>>> find the offending line.
>>>
>>> Regards
>>> Franz
>>>
>>> Am Fr., 3. Jan. 2020 um 11:44 Uhr schrieb Mickaël Bucas <
>>> mbu...@gmail.com>:
>>>
>>>> Hello
>>>>
>>>> For some time I've been wondering why I had a difference on
>>>> dev-python/olefile-0.46 between 2 machines : one was installed with
>>>> python_targets_python3_7, the other wasn't.
>>>> And I finally pinpointed it to package.accept_keywords containing
>>>> "dev-python/olefile ~amd64" on one of the machines only
>>>>
>>>> At the time of writing, dev-python/olefile-0.46 is the stable version,
>>>> and KEYWORDS contains "amd64" (no tilde) among others.
>>>>
>>>> When package.accept_keywords doesn't contain "dev-python/olefile
>>>> ~amd64", I get :
>>>> emerge -pv1 --verbose-conflicts olefile
>>>> These are the packages that would be merged, in order:
>>>> Calculating dependencies... done!
>>>> [ebuild   R] dev-python/olefile-0.46::gentoo  USE="-doc"
>>>> PYTHON_TARGETS="python2_7 python3_6 (-pypy3) (-python3_7) (-python3_8)" 0
>>>> KiB
>>>> Total: 1 package (1 reinstall), Size of downloads: 0 KiB
>>>>
>>>> => Python 3.7 is disabled
>>>>
>>>> When package.accept_keywords contains "dev-python/olefile ~amd64", I
>>>> get :
>>>> emerge -pv1 olefile
>>>> These are the packages that would be merged, in order:
>>>> Calculating dependencies... done!
>>>> [ebuild   R] dev-python/olefile-0.46::gentoo  USE="-doc"
>>>> PYTHON_TARGETS="python2_7 python3_6 python3_7* -pypy3 -python3_8" 0 KiB
>>>> Total: 1 package (1 reinstall), Size of downloads: 0 KiB
>>>>
>>>> => Python 3.7 is enabled
>>>>
>>>> It seems really really strange to me for the same version of a stable
>>>> package to be "influenced" by keywording.
>>>> Is it a bug or a feature ?
>>>> Did I do something wrong ?
>>>>
>>>> Thanks
>>>> Best regards
>>>> Mickaël Bucas
>>>>
>>>

Re: [gentoo-user] Stable Python package changes USE flags with ~amd64

[Mickaël Bucas]

Hi Franz

Thanks for your reply.

However your assumption is incorrect: these two commands are run on the
same machine, with only the keyword on "olefile" changed.
Thinking a bit more about it, Python 3.7 isn't stable yet, so I also have
"=dev-lang/python-3.7* ~amd64" in package.accept_keyword.

I've been able to reproduce this behavior in a chroot based on stage 3 with
the minimum packages installed.
I have in make.conf
PYTHON_TARGETS="python2_7 python3_6 python3_7"
In /var/lib/portage/world
dev-lang/python:3.7
dev-python/olefile
In /etc/portage/package.accept_keywords
dev-python/olefile ~amd64
=dev-lang/python-3.7* ~amd64
dev-python/setuptools ~amd64
dev-python/certifi ~amd64

And emerge says :
# emerge -pv1 olefile
These are the packages that would be merged, in order:
Calculating dependencies... done!
[ebuild   R] dev-python/olefile-0.46::gentoo  USE="-doc"
PYTHON_TARGETS="python2_7 python3_6 python3_7 -pypy3 -python3_8" 0 KiB
Total: 1 package (1 reinstall), Size of downloads: 0 KiB

When I remove " dev-python/olefile ~amd64", Python 3.7 would be disabled :
# emerge -pv1 olefile
These are the packages that would be merged, in order:
Calculating dependencies... done!
[ebuild   R] dev-python/olefile-0.46::gentoo  USE="-doc"
PYTHON_TARGETS="python2_7 python3_6 (-pypy3) (-python3_7*) (-python3_8)" 0
KiB
Total: 1 package (1 reinstall), Size of downloads: 0 KiB

This is still puzzling me, but one interpretation may be :
I you enable the unstable ~amd64 keyword on a package, the stable version
of said package is allowed to run on the unstable version of the Python
interpreter.

This seems to be the intended behavior, as I found that at least 40 Python
packages on each of my 2 systems are stable and have Python 3.7 enabled (I
keyworded all of them sometime in the past...)

Thanks
Best regards
Mickaël Bucas

Le mar. 7 janv. 2020 à 08:08, Franz Fellner  a
écrit :

> I assume those emerge commands weren't done on one machine but come from
> those two different machines.
> This change in USE Flags can't come from that line in
> package.accept_keywords.
> This is a change in PYTHON_TARGETS in make.conf, package.use or
> package.env.
> Carefully go through those config files/directories, I am sure you will
> find the offending line.
>
> Regards
> Franz
>
> Am Fr., 3. Jan. 2020 um 11:44 Uhr schrieb Mickaël Bucas  >:
>
>> Hello
>>
>> For some time I've been wondering why I had a difference on
>> dev-python/olefile-0.46 between 2 machines : one was installed with
>> python_targets_python3_7, the other wasn't.
>> And I finally pinpointed it to package.accept_keywords containing
>> "dev-python/olefile ~amd64" on one of the machines only
>>
>> At the time of writing, dev-python/olefile-0.46 is the stable version,
>> and KEYWORDS contains "amd64" (no tilde) among others.
>>
>> When package.accept_keywords doesn't contain "dev-python/olefile ~amd64",
>> I get :
>> emerge -pv1 --verbose-conflicts olefile
>> These are the packages that would be merged, in order:
>> Calculating dependencies... done!
>> [ebuild   R] dev-python/olefile-0.46::gentoo  USE="-doc"
>> PYTHON_TARGETS="python2_7 python3_6 (-pypy3) (-python3_7) (-python3_8)" 0
>> KiB
>> Total: 1 package (1 reinstall), Size of downloads: 0 KiB
>>
>> => Python 3.7 is disabled
>>
>> When package.accept_keywords contains "dev-python/olefile ~amd64", I get
>> :
>> emerge -pv1 olefile
>> These are the packages that would be merged, in order:
>> Calculating dependencies... done!
>> [ebuild   R] dev-python/olefile-0.46::gentoo  USE="-doc"
>> PYTHON_TARGETS="python2_7 python3_6 python3_7* -pypy3 -python3_8" 0 KiB
>> Total: 1 package (1 reinstall), Size of downloads: 0 KiB
>>
>> => Python 3.7 is enabled
>>
>> It seems really really strange to me for the same version of a stable
>> package to be "influenced" by keywording.
>> Is it a bug or a feature ?
>> Did I do something wrong ?
>>
>> Thanks
>> Best regards
>> Mickaël Bucas
>>
>

[gentoo-user] Stable Python package changes USE flags with ~amd64

[Mickaël Bucas]

Hello

For some time I've been wondering why I had a difference on
dev-python/olefile-0.46 between 2 machines : one was installed with
python_targets_python3_7, the other wasn't.
And I finally pinpointed it to package.accept_keywords containing
"dev-python/olefile ~amd64" on one of the machines only

At the time of writing, dev-python/olefile-0.46 is the stable version, and
KEYWORDS contains "amd64" (no tilde) among others.

When package.accept_keywords doesn't contain "dev-python/olefile ~amd64", I
get :
emerge -pv1 --verbose-conflicts olefile
These are the packages that would be merged, in order:
Calculating dependencies... done!
[ebuild   R] dev-python/olefile-0.46::gentoo  USE="-doc"
PYTHON_TARGETS="python2_7 python3_6 (-pypy3) (-python3_7) (-python3_8)" 0
KiB
Total: 1 package (1 reinstall), Size of downloads: 0 KiB

=> Python 3.7 is disabled

When package.accept_keywords contains "dev-python/olefile ~amd64", I get :
emerge -pv1 olefile
These are the packages that would be merged, in order:
Calculating dependencies... done!
[ebuild   R] dev-python/olefile-0.46::gentoo  USE="-doc"
PYTHON_TARGETS="python2_7 python3_6 python3_7* -pypy3 -python3_8" 0 KiB
Total: 1 package (1 reinstall), Size of downloads: 0 KiB

=> Python 3.7 is enabled

It seems really really strange to me for the same version of a stable
package to be "influenced" by keywording.
Is it a bug or a feature ?
Did I do something wrong ?

Thanks
Best regards
Mickaël Bucas

Re: [gentoo-user] Noah's ArK

[Mickaël Bucas]

Hi

If it's only for Python 2 packages, it's easy with "virtualenv", I've
done it for multiple independent Trac installations, each with a
different set of plugins. You still have to keep the Python 2 base
packages on the system.

If it's for 32 bits packages, you can isolate them in a chroot,
following the Gentoo guide [1]. I remember having used this to install
binary packages like Adobe Acrobat when multilib was supported only
with 32bits binary packages (I don't remember what made it necessary).
If the chroot is mounted at system start, you can define menu actions
in your desktop environment to launch commands inside of the chroot.
You may need to bind mount your home directory or other data
directories inside of the chroot to make them visible to these
programs.

The chroot solution can also work for Python 2, and it's not limited
to 32 bits : you can create the chroot with a 64 bits stage 3 tarball.

Best regards

Mickaël Bucas
[1] https://wiki.gentoo.org/wiki/Project:AMD64/32-bit_Chroot_Guide

Le dim. 17 nov. 2019 à 12:20, Helmut Jarausch  a écrit :
>
> Hi,
>
> I'd like to "preserve" some packages which do require components I
> don't like to have "regularly" installed any more like some depending
> on Python2 or are 32bit.
>
> Is there a means to build a "mini binary system" where I can put these?
>
> Many thanks for a hint,
> Helmut

Re: [gentoo-user] where is /usr/portage?

[Mickaël Bucas]

Le lun. 11 nov. 2019 à 14:38, Dale  a écrit :
>
> Mick wrote:
> > On Monday, 11 November 2019 13:00:20 GMT Dale wrote:
> >> Mickaël Bucas wrote:
> >>> Le lun. 11 nov. 2019 à 09:35, Mick  a écrit :
> >>>> On Monday, 11 November 2019 08:25:06 GMT n952162 wrote:
> >>>>> I re-installed gentoo from the minimal boot cd (amd64), re-emerged
> >>>>> everything from my old, saved world file, overnight, and its up and
> >>>>> running, more or less.  Then, I wanted to see what was available and
> >>>>> discovered, there's no /usr/portage directory! What did I do wrong?
> >>>> Nothing.
> >>>>
> >>>> /usr/portage has now moved to /var/db/repos/gentoo/.
> >>>>
> >>>> /usr/portage/distfiles has moved to /var/cache/distfiles/
> >>>>
> >>>> Portage will work fine with both legacy and new fs locations.
> >>>> --
> >>>> Regards,
> >>>>
> >>>> Mick
> >>> My two systems are currently using the old locations.
> >>> Is there a documentation about the way to migrate to the new locations
> >>> without breaking things ?
> >>> The profile links comes to mind but other things are probably necessary !
> >>>
> >>> Has anyone already done the migration ?
> >>> In this case do you have advices or warnings about it ?
> >>>
> >>> Thanks
> >>>
> >>> Best regards
> >>> Mickaël Bucas
> >> Others have posted some good info but sometimes a example that is in use
> >> can help a lot.  Here's what is in my make.conf:
> >>
> >>
> >> root@fireball / # cat /etc/make.conf | grep var
> >> PORT_LOGDIR=/var/log/portage/
> >> source /var/lib/layman/make.conf
> >> DISTDIR="/var/cache/portage/distfiles/"
> >> PKGDIR="/var/cache/portage/packages"
> >> PORTDIR="/var/cache/portage/tree"
> >> root@fireball / #
> >>
> >>
> >> The last three are what you need to look at.
> > ... AND ...
> >
> > the last thing (PORTDIR) is what you should no longer have specified in 
> > /etc/
> > portage/make.conf, but in /etc/portage/repos.conf/gentoo.conf:
> >
> > $ grep location /etc/portage/repos.conf/gentoo.conf
> > #location = /usr/portage  <==legacy portage fs location
> > location = /var/db/repos/gentoo
> >
> > HTH.
>
>
> A, that one line is in there so I guess it got moved at some point,
> likely the devs had some magic going on and did it for me.  ;-)  I'll
> comment that out in make.conf, so I don't confuse myself later on.  :/
> Come to think of it, I'll add a comment as to where it moved to as well
> so I don't have to go dig for it some day.
>
> Thanks.  This will help the OP as well.
>
> Dale
>
> :-)  :-)
>
Thanks for all explanations.
I've moved the Gentoo tree and distfiles from /usr to /var.
Having these in /usr has always felt strange to me but I never thought
it was so easy to move them !

Best regards
Mickaël Bucas

Re: [gentoo-user] where is /usr/portage?

[Mickaël Bucas]

Le lun. 11 nov. 2019 à 09:35, Mick  a écrit :
>
> On Monday, 11 November 2019 08:25:06 GMT n952162 wrote:
> > I re-installed gentoo from the minimal boot cd (amd64), re-emerged
> > everything from my old, saved world file, overnight, and its up and
> > running, more or less.  Then, I wanted to see what was available and
> > discovered, there's no /usr/portage directory! What did I do wrong?
>
> Nothing.
>
> /usr/portage has now moved to /var/db/repos/gentoo/.
>
> /usr/portage/distfiles has moved to /var/cache/distfiles/
>
> Portage will work fine with both legacy and new fs locations.
> --
> Regards,
>
> Mick

My two systems are currently using the old locations.
Is there a documentation about the way to migrate to the new locations
without breaking things ?
The profile links comes to mind but other things are probably necessary !

Has anyone already done the migration ?
In this case do you have advices or warnings about it ?

Thanks

Best regards
Mickaël Bucas

Re: [gentoo-user] almost free launch: an idea to lower build time, and rice, at the same time

[Mickaël Bucas]

lower minimum popularity than those small
>   packages.  Choosing the sweet-spot trade-off is a matter of optimizing
>   resources of the central server.
>
> * The statistical analysis in steps (6) to (8) could also be further enhanced
>   by ranking individual users who upload the binaries.  Users, who upload 
> bins,
>   could optionally also sign their packages, and henceforth be identified by
>   the central server.  Eventually, statistics can be used to also calculate a
>   confidence measure on how trusty a user is.  This can eventually help the
>   server more accurately calculate the confidence of the uploaded bins, by 
> also
>   incorporating the past history of those users.
>
>   Sub-note 1:  The reason signing is optional, is because ---thanks to
>   information theory--- we don't really need signed packages in order to know
>   that a package is not an outlier.  I.e. even unsigned packages can help us
>   figure out the probability of error by simply looking at the redundancy
>   counts.
>
>   Sub-note 2:  But, of course, signing would help as it will allow the central
>   server's statistical analysis to also put into account which bin is coming
>   from which user.  E.g. not all users are equally trusty, and this can help
>   the system be more accurate in its prediction of the error on the package.
>
>   Sub-note 3:  I said it already, but just to repeat, when the error becomes
>   low enough, this distributed system can potentially end up producing 
> binaries
>   that match or exceed trusty Gentoo devs.  Adding common heuristic checks are
>   optional, but can make the bins even more likely to beat manual devs.
>
> * Eventually, this statistical approach could also replace the need for
>   manually electing binary package maintainers by a principled statistical
>   approach.  Thanks to the way stuff work in nature, this system has the
>   potential of being even more trusty than the trustier bin-packager 
> developer.
>
> * In the future, this could be extended to source-code ebuilds, too.
>   Ultimately, reaching a quality equal to, or exceeding that of, the current
>   manual system.  This may pave the path to a much more efficient operating
>   system where less manual labour is needed by the devs, so that more devs can
>   do actually more fun things than packaging boring stuff.
>
> * This system will get better the more people use it, and the better it gets
>   the more the people would like it and hence even more will use it!  It works
>   like turbo-charging.  Hence, if this succeeds, we may market Gentoo as the
>   first "turbo-charged OS"!
>
> * Based on step (5), the server can set frequency thresholds in order to keep
>   its resources only utilized by highly demanded packages.
>
>
> rgrds,
> cm

Hi Caveman

The Portage tree contains a few binary packages prepared by Gentoo
developers, like Firefox, Rust, LibreOffice...
"ls -d /usr/portage/*/*-bin" shows about 90 packages prepared in this
way, some of them because they are non-free like Oracle JDK

This means that there is no necessary changes to Gentoo to accomplish
what you describe : compile the packages, write the ebuilds for the
binary packages, publish ebuilds in an overlay.

But the really short list above shows that it's a really complex task
because of all dependencies and configurable elements in Gentoo. If
you just have a look at the output of "emerge --info" you can imagine
all the moving parts, like compiler versions and compile options,
Bash, Perl, Python, Init system, USE flags (combinatorial), even human
languages. And that is just the easily visible parts !

I remember reading an article about a man trying to reproduce binary
packages of a binary distribution and failing to do so, because there
are so many parts involved. I've read later that distributions have
done some work to have reproducible builds, but I'm not sure how
successful they are, even when all choices are predefined.

Given that Gentoo has taken a whole different road by having more
choices available to the user, I don't think the compilation results
of one configuration would be easily used on another.

To go even further, pushing your compiled packages to a public server
may create a security risk by exposing many parts of your
configuration that could be analyzed by malicious people.

So far I don't see a really big advantage in building this kind of
infrastructure compared to either a binary distribution or Gentoo with
home compilation.

Best regards

Mickaël Bucas

Re: [gentoo-user] Question : Fixing @x11-module-rebuild

[Mickaël Bucas]

Le mer. 18 sept. 2019 à 04:31, Corbin  a écrit :
>
>
> On 9/17/19 2:41 PM, Jack wrote:
> > I have one older box which still behaves "correctly."  On my newer box,
> > built about a month ago, I get the same behavior you describe.  I have
> > no explanation.
> >
> > Jack
>
> Looks like an emerge bug. :(
> Hoped for an easy fix.
On my two machines, the problem doesn't happen. They are old, but I
can't imagine what would be the difference with a new installation.

>
> Well, I'll file a bug report with Gentoo first thing in the morning.
>
> Corbin
>
It seems someone already did it :
https://bugs.gentoo.org/693980 - sys-apps/portage -
@x11-module-rebuild doesn't work for amd64
The good news is that the fix is done, and will be available soon.

Best regards

Mickaël Bucas

Re: [gentoo-user] Emerge wants to downgrade icu

[Mickaël Bucas]

.0.4:0/3::gentoo, ebuild scheduled for merge)
> (dev-lang/spidermonkey-52.9.1_pre1:52/52::gentoo, ebuild scheduled for 
> merge)
> (media-libs/harfbuzz-2.3.1:0/0.9.18::gentoo, ebuild scheduled for merge)
> (dev-tex/bibtexu-3.71_p20170524:0/0::gentoo, ebuild scheduled for merge)
> (app-text/libqxp-0.0.2:0/0::gentoo, ebuild scheduled for merge)
> (media-libs/raptor-2.0.15-r2:2/2::gentoo, ebuild scheduled for merge)
>
> The following mask changes are necessary to proceed:
>  (see "package.unmask" in the portage(5) man page for more details)
> # required by @__auto_slot_operator_replace_installed__ (argument)
> # /usr/portage/profiles/base/package.mask:
> # Andreas Sturmlechner  (15 Jul 2018)
> # Old ICU is unsupported. ICU 58 only remains for 13.0 based profiles.
> =dev-libs/icu-58.2-r1
>
> NOTE: The --autounmask-keep-masks option will prevent emerge
>   from creating package.unmask or ** keyword changes.
>
> Would you like to add these changes to your config files? [Yes/No] No
>
> --
> Alarig
>

When I met this situation, I had to add keyword ~amd64 for
app-office/libreoffice-bin
Then emerge finds a better solution : upgrade icu to 64.2

Cordialement

Mickaël Bucas

Re: [gentoo-user] Re: Incorrect list of groups membership

[Mickaël Bucas]

Hi

I've opened issue #1159 on SDDM Github [1], with a patch that makes
better use of "initgroups" and "pam_setcreds", I hope.

@Remy Blank if you could try the patch and comment on the issue, that
may help SDDM project members to evaluate it.

Thanks

Best regards
Mickaël Bucas

[1] https://github.com/sddm/sddm/issues/1159

Le ven. 19 avr. 2019 à 00:34, Mickaël Bucas  a écrit :
>
> Le lun. 15 avr. 2019 à 20:26, Remy Blank  a écrit :
> >
> > > After a reboot, the problem disappears for a while, but comes again,
> > > and I didn't find what could trigger it.
> > > I can't figure what KDE could have to do with user groups returned by
> > > the kernel !
> > >
> > > Does anyone have a hint on the origin of this problem ?
> >
> > Yes, this is triggered by restarting the xdm service, possibly limited
> > to sddm users.
> >
> > I have noticed the same issue here. Groups are correct after a reboot,
> > but if I do:
> >
> > $ /etc/init.d/xdm restart
> >
> > and log into KDE, then I'm a member of all sorts of system groups. I'm
> > using sddm, maybe it happens with other login managers as well.
> >
> > I suspect that this is due to inheriting the supplementary groups of
> > which "root" is a member at the time the login manager is started. At
> > boot time, it is a member of no additional groups, whereas in a root
> > shell, it is:
> >
> > # groups
> > root bin daemon sys adm disk wheel floppy dialout tape video
> >
> > I suspect this is a bug in sddm, or maybe in pam. It should drop all
> > supplementary groups before switching to the user logging in.
> >
> > As a workaround, I now always reboot instead of restarting xdm.
> >
> > -- Remy
>
> Thanks for pointing me to SDDM. After looking at the source code [1]
> that prepares the user session I think it's not correct.
> The main problem is that it doesn't call "initgroups" when PAM is
> enabled. This function is the one that loads the list of local groups
> for the user.
> PAM functions should be called afterwards to load additional groups
> with for example pam_group.so, according to man pam_setcred(3)
>
> This is confirmed by looking at the code of XDM for the same action,
> which calls "initgroups" in all cases [2].
> I found this piece of code interesting with for example :
> # if defined(BSD) && (BSD >= 199103)
> So it seems it's been around for a really long time !
>
> I also found issue 416 on SDDM [3] on Github that seems to be the same
> problem, but it has been closed, with a dubious explanation that this
> would be caused by KDE kinit.
> I don't believe it because KDE processes are launched with the user
> session id, and therefore shouldn't be able to add system groups to
> the session.
> And if I look at the groups of sddm-greeter when it's running under
> user "sddm", they are already incorrect before a KDE session is
> opened.
> mick@xxx ~ $ grep Groups /proc/$(pidof sddm-greeter)/status
> Groups: 0 1 2 3 4 6 10 11 26 27 27 243
>
> I will try to modify SDDM code to include "initgroups" where I suppose
> it should be called.
> This could result in a new bug or a pull request depending on the results...
>
> Best regards
> Mickaël Bucas
>
> [1] https://github.com/sddm/sddm/blob/develop/src/helper/UserSession.cpp
> [2] https://gitlab.freedesktop.org/xorg/app/xdm/blob/master/xdm/session.c
> [3] https://github.com/sddm/sddm/issues/416

Re: [gentoo-user] Re: Incorrect list of groups membership

[Mickaël Bucas]

Le lun. 15 avr. 2019 à 20:26, Remy Blank  a écrit :
>
> > After a reboot, the problem disappears for a while, but comes again,
> > and I didn't find what could trigger it.
> > I can't figure what KDE could have to do with user groups returned by
> > the kernel !
> >
> > Does anyone have a hint on the origin of this problem ?
>
> Yes, this is triggered by restarting the xdm service, possibly limited
> to sddm users.
>
> I have noticed the same issue here. Groups are correct after a reboot,
> but if I do:
>
> $ /etc/init.d/xdm restart
>
> and log into KDE, then I'm a member of all sorts of system groups. I'm
> using sddm, maybe it happens with other login managers as well.
>
> I suspect that this is due to inheriting the supplementary groups of
> which "root" is a member at the time the login manager is started. At
> boot time, it is a member of no additional groups, whereas in a root
> shell, it is:
>
> # groups
> root bin daemon sys adm disk wheel floppy dialout tape video
>
> I suspect this is a bug in sddm, or maybe in pam. It should drop all
> supplementary groups before switching to the user logging in.
>
> As a workaround, I now always reboot instead of restarting xdm.
>
> -- Remy

Thanks for pointing me to SDDM. After looking at the source code [1]
that prepares the user session I think it's not correct.
The main problem is that it doesn't call "initgroups" when PAM is
enabled. This function is the one that loads the list of local groups
for the user.
PAM functions should be called afterwards to load additional groups
with for example pam_group.so, according to man pam_setcred(3)

This is confirmed by looking at the code of XDM for the same action,
which calls "initgroups" in all cases [2].
I found this piece of code interesting with for example :
# if defined(BSD) && (BSD >= 199103)
So it seems it's been around for a really long time !

I also found issue 416 on SDDM [3] on Github that seems to be the same
problem, but it has been closed, with a dubious explanation that this
would be caused by KDE kinit.
I don't believe it because KDE processes are launched with the user
session id, and therefore shouldn't be able to add system groups to
the session.
And if I look at the groups of sddm-greeter when it's running under
user "sddm", they are already incorrect before a KDE session is
opened.
mick@xxx ~ $ grep Groups /proc/$(pidof sddm-greeter)/status
Groups: 0 1 2 3 4 6 10 11 26 27 27 243

I will try to modify SDDM code to include "initgroups" where I suppose
it should be called.
This could result in a new bug or a pull request depending on the results...

Best regards
Mickaël Bucas

[1] https://github.com/sddm/sddm/blob/develop/src/helper/UserSession.cpp
[2] https://gitlab.freedesktop.org/xorg/app/xdm/blob/master/xdm/session.c
[3] https://github.com/sddm/sddm/issues/416

[gentoo-user] Incorrect list of groups membership

[Mickaël Bucas]

Hi,

On my 2 Gentoo machines, users have a strange problem of group
membership. When using 'id' or 'groups' without arguments, the list of
groups includes those from the root user, and groups common to the
user and root are duplicated, like wheel, video, allowssh.
I observed that the list of groups is incorrect only in a KDE session,
either in Konsole or xterm, but the list of groups is correct in other
kinds of sessions like on a TTY or through SSH.
After a reboot, the problem disappears for a while, but comes again,
and I didn't find what could trigger it.
I can't figure what KDE could have to do with user groups returned by
the kernel !

Does anyone have a hint on the origin of this problem ?

mick@xxx ~ $ groups
root bin daemon sys adm disk wheel wheel floppy uucp cron audio cdrom
dialout tape video video games cdrw apache usb vboxusers portage
allowssh allowssh svn users mick
mick@xxx ~ $ id
uid=1001(mick) gid=1001(mick)
groupes=1001(mick),0(root),1(bin),2(daemon),3(sys),4(adm),6(disk),10(wheel),11(floppy),14(uucp),16(cron),18(audio),19(cdrom),20(dialout),26(tape),27(video),35(games),80(cdrw),81(apache),85(usb),102(vboxusers),250(portage),800(allowssh),909(svn),1000(users)

When run with a login, the list is correct

mick@xxx ~ $ groups mick
wheel cron audio cdrom video games cdrw apache usb vboxusers portage
allowssh svn users mick
mick@xxx ~ $ groups root
root bin daemon sys adm disk wheel floppy uucp dialout tape video allowssh
mick@xxx ~ $ id mick
uid=1001(mick) gid=1001(mick)
groupes=1001(mick),10(wheel),16(cron),18(audio),19(cdrom),27(video),35(games),80(cdrw),81(apache),85(usb),102(vboxusers),250(portage),800(allowssh),909(svn),1000(users)
mick@xxx ~ $ id root
uid=0(root) gid=0(root)
groupes=0(root),1(bin),2(daemon),3(sys),4(adm),6(disk),10(wheel),11(floppy),14(uucp),20(dialout),26(tape),27(video),800(allowssh)

As far as I can tell the contents of '/etc/passwd' and '/etc/group' is
also correct.

mick@xxx ~ $ egrep 'mick|root' /etc/passwd
root:x:0:0:root:/root:/bin/bash
operator:x:11:0:operator:/root:/bin/bash
mick:x:1001:1001::/home/mick:/bin/bash

mick@xxx ~ $ egrep 'mick|root' /etc/group
root:x:0:root
bin:x:1:root,bin,daemon
daemon:x:2:root,bin,daemon
sys:x:3:root,bin,adm
adm:x:4:root,adm,daemon
disk:x:6:root,adm,haldaemon
wheel:x:10:root,mick,jef,apache,anne
floppy:x:11:root,haldaemon
uucp:x:14:root
cron:x:16:cron,mick,apache
audio:x:18:famille,mick,jef,juliette,victor,anne,pulse,sddm
cdrom:x:19:famille,mick,haldaemon,jef,juliette,victor,anne
dialout:x:20:root
tape:x:26:root
video:x:27:root,famille,mick,jef,juliette,victor,anne,oracle,sddm
games:x:35:famille,mick,jef,juliette,victor,anne
cdrw:x:80:famille,mick,haldaemon
apache:x:81:famille,jef,mick
usb:x:85:famille,mick,haldaemon,juliette,victor,anne
vboxusers:x:102:famille,vbox,mick,jef
portage:x:250:portage,famille,mick,jef,apache
allowssh:x:800:mick,jef,root,anne,juliette,victor
svn:x:909:famille,jef,mick,tracd
users:x:1000:mick,jef,apache,juliette,victor,offlineimap,anne
mick:x:1001:mick

The difference in output between 'id' and 'id mick' happens because
'id' calls the syscall 'getgroups' in the first case, but not in the
other, as I could see with 'strace'

mick@xxx $ strace id
[...]
getgroups(0, NULL)  = 29
getgroups(29, [0, 1, 2, 3, 4, 6, 10, 10, 11, 14, 16, 18, 19, 20, 26,
27, 27, 33, 35, 80, 81, 85, 102, 250, 800, 800, 909, 1000, 1001]) = 29
[...]

mick@xxx $ strace id mick
[...]
openat(AT_FDCWD, "/var/db/group.db", O_RDONLY|O_CLOEXEC) = -1 ENOENT
(Aucun fichier ou dossier de ce type)
openat(AT_FDCWD, "/etc/group", O_RDONLY|O_CLOEXEC) = 3
lseek(3, 0, SEEK_CUR)   = 0
fstat(3, {st_mode=S_IFREG|0644, st_size=1978, ...}) = 0
read(3, "root:x:0:root\nbin:x:1:root,bin,d"..., 4096) = 1978
lseek(3, 0, SEEK_CUR)   = 1978
[...repeated]
lseek(3, 0, SEEK_CUR)   = 1978
read(3, "", 4096)   = 0
close(3)= 0
openat(AT_FDCWD, "/var/db/group.db", O_RDONLY|O_CLOEXEC) = -1 ENOENT
(Aucun fichier ou dossier de ce type)
openat(AT_FDCWD, "/etc/group", O_RDONLY|O_CLOEXEC) = 3
lseek(3, 0, SEEK_CUR)   = 0
fstat(3, {st_mode=S_IFREG|0644, st_size=1978, ...}) = 0
read(3, "root:x:0:root\nbin:x:1:root,bin,d"..., 4096) = 1978
lseek(3, 0, SEEK_CUR)   = 1978
[...repeated]
lseek(3, 0, SEEK_CUR)   = 1978
read(3, "", 4096)   = 0
close(3)= 0
[...]

mick@xxx ~ # uname -a
Linux xxx 4.19.27-gentoo-r1 #1 SMP Mon Apr 1 14:38:01 CEST 2019 x86_64
Intel(R) Celeron(R) CPU G1610T @ 2.30GHz GenuineIntel GNU/Linux

Thanks

Best regards
Mickaël Bucas

Re: [gentoo-user] nx / nxclient - replacement

[Mickaël Bucas]

2017-02-06 10:55 GMT+01:00 Helmut Jarausch :
> On 02/06/2017 08:53:19 AM, the...@sys-concept.com wrote:
>>
>> Are there any good replacement for "nx / nxclient" in Linux?
>> NX is long time gone from portage.  I hope, I can still install them
>> from atic.
>> This was another reason I wasn't upgrading for a long time as I need
>> them to access remote boxes in GUI.
>>
>
> I have net-misc/x2goclient (and net-misc/x2goserver) installed.
> But I think, x2goserver has to be installed on the remote system.

There is also Guacamole, whose main advantage is that the client is a browser.
The architecture seems a bit complex.
I intend to try it but I can tell currently how good it is.
http://guacamole.incubator.apache.org/
It's masked in Portage :
net-misc/guacamole-server
www-apps/guacamole

Best regards
Mickaël

Re: [gentoo-user] unexpected problems when switching c compilers

[Mickaël Bucas]

2016-10-25 20:02 GMT+02:00 Miroslav Rovis <miro.ro...@croatiafidelis.hr>:

> On 161025-12:02-0400, John Covici wrote:
> > Hi.  I finally decided to switch the c compiler from 4.9.3 to 5.3.0
> > and ran into some strange problems.  Although the kernel compiled
> > successfully, several packages in my last update of world got
> > undefined references, one of which was mktoolnix and also mpv.  Now I
> > did solve them by re-emerging the packages containing the references
> > and that fix things, although I had to do it in stages, getting past
> > one and finding another one.
> >
> > So, my question is, is there away to fix this throuought /usr/lib
> > without having to re-emerge world or something, or do I just wait and
> > take them one at a time during future updates?
> >
> > Thanks in advance for any suggestions.
>
> I've been using the 5.x since long months, since I'm on ~amd64. I
> remember vaguely there was an item in the eselect news about it, and
> links there and things. It should still be in your news list.
>

There was probably an Application Binary Interface change between the 2
compilers, for example [1]
The news from portage [2] explain how to use revdep-rebuild

All I know about these problems, I've learn from Diego Petteno's blog [3]
and the articles about Application Binary Interface
I've myself encountered this kind of problem even between minor versions of
GCC when the C++11 ABI was not stable.

I think you have some happy compiling ahead !

Best regards
Mickaël Bucas

[1] https://gcc.gnu.org/onlinedocs/libstdc%2B%2B/manual/using_dual_abi.html
[2]
https://www.gentoo.org/support/news-items/2015-10-22-gcc-5-new-c++11-abi.html
[3] https://blog.flameeyes.eu/2010/09/your-worst-enemy-undefined-symbols/

Re: [gentoo-user] Missing digest for *** Tree looks messed up.

[Mickaël Bucas]

2015-08-10 2:15 GMT+02:00 Dale rdalek1...@gmail.com:

 Howdy,

 I'm sure I'm not alone on monitoring -dev to see upcoming changes.  I
 noticed they switched to git or something and it seems to have caused a
 bit of trouble.  At least I think it may have.  I did my usual 'eix-sync
  emerge -uvaDN world'.  The sync took MUCH longer than usual.  I'm
 talking a WHOLE LOT longer than usual.  My first thought, one time thing
 because of the changes, maybe.  Then I got a screen full of this sort of
 stuff.



  * Missing digest for

'/var/cache/portage/tree/media-libs/fontconfig/fontconfig-2.11.1-r2.ebuild'
[...]

I got a similar list about all my installed packages. The reason is that
the Manifest files have changed.
Before there was a digest for each ebuild in a category, as can be seen
here :
https://github.com/gentoo/gentoo-portage-rsync-mirror/blob/master/kde-base/knotes/Manifest
Now there is only the digests for source files
https://gitweb.gentoo.org/repo/gentoo.git/tree/kde-base/knotes/Manifest


 Can I take that just saying above back and say FUBAR instead?   :-(

 Thoughts??

 Dale

Either the Manifest generator encountered a problem, or the digest for
ebuilds is not necessary in new versions of portage, but the latter is less
likely in my opinion as it would lower overall security.
I didn't find a bug about this subject, but I don't really know what and
where to search...

Who could we ask for more insight into that strange behavior ?

Best regards
Mickaël Bucas

Re: [gentoo-user] Missing digest for *** Tree looks messed up.

[Mickaël Bucas]

2015-08-10 10:55 GMT+02:00 Mickaël Bucas mbu...@gmail.com:


 2015-08-10 2:15 GMT+02:00 Dale rdalek1...@gmail.com:
 
  Howdy,
 
  I'm sure I'm not alone on monitoring -dev to see upcoming changes.  I
  noticed they switched to git or something and it seems to have caused a
  bit of trouble.  At least I think it may have.  I did my usual 'eix-sync
   emerge -uvaDN world'.  The sync took MUCH longer than usual.  I'm
  talking a WHOLE LOT longer than usual.  My first thought, one time thing
  because of the changes, maybe.  Then I got a screen full of this sort of
  stuff.
 
 
 
   * Missing digest for
 
'/var/cache/portage/tree/media-libs/fontconfig/fontconfig-2.11.1-r2.ebuild'
 [...]

 I got a similar list about all my installed packages. The reason is that
the Manifest files have changed.
 Before there was a digest for each ebuild in a category, as can be seen
here :

https://github.com/gentoo/gentoo-portage-rsync-mirror/blob/master/kde-base/knotes/Manifest
 Now there is only the digests for source files
 https://gitweb.gentoo.org/repo/gentoo.git/tree/kde-base/knotes/Manifest

 
  Can I take that just saying above back and say FUBAR instead?   :-(
 
  Thoughts??
 
  Dale

 Either the Manifest generator encountered a problem, or the digest for
ebuilds is not necessary in new versions of portage, but the latter is less
likely in my opinion as it would lower overall security.
 I didn't find a bug about this subject, but I don't really know what and
where to search...

So the answer was in Git history, in the first commit :
https://gitweb.gentoo.org/repo/gentoo.git/log/kde-base/knotes/Manifest?showmsg=1

3. Transform all Manifests to thin

https://wiki.gentoo.org/wiki/Repository_format/package/Manifest
A Thin Manifest is a Manifest file in which checksums are stored only for
distfiles (*DIST* type) and not for files inside the repository. The
motivation for that is whenever the repository is fetched through a VCS
which ensures local file integrity already.

I think I understand the problem a bit better : the CVS repository
contained classic Manifest files, and was exposed directly to rsync. The
Git repository contains only thin Manifest, but has been set as the source
for rsync at one point !

This has already been reported, and solved.
https://bugs.gentoo.org/show_bug.cgi?id=557184

Best regards
Mickaël Bucas

Re: [gentoo-user] [OT] Strange behaviour of google certificates.

[Mickaël Bucas]

2015-04-01 19:19 GMT+02:00 Gevisz gev...@gmail.com:
 This question does specifically relates to Gentoo distribution
 but, as far as I have not subscribed to any other mailing list,
 I dare to ask it here.

 So, I am using Claws Mail that downloads e-mails from several
 google mail accounts (all are mine :) and about once or twice
 in a month get into the situation when Claws asks me to verify
 and change the google certificates, first in one direction and
 soon after that (usually during the next downloading of my e-mails)
 - in another.

 The situation is illustrated by the 2 message screenshots that are
 attached to this e-mail.

 The strange thing for me is that, first, the Claws asks me to verify
 and accept a newer certificate complaing that the old one is in some
 aspect bad, and soon after that it complains about a newer certificate
 and asks me to verify and and accept the older one.

 I suspect that it is google that makes something wrong here.

 What do you think?

Hi Gevisz

I had a similar behavior with another tools : offlineimap
It seems that Google changes certificates very often and/or uses
different certificates on different connections

For offlineimap, the solution is to use an option to check certificates :
sslcacertfile = /etc/ssl/certs/ca-certificates.crt

Maybe there is an option to do the same in Claws Mail.
I found Bug 2199 - Claws doesn't propery verify certification chain
[1] which affected a GMail user.
It's fixed, so you may find what's been done.

Best regards

Mickaël Bucas

[1] http://www.thewildbeast.co.uk/claws-mail/bugzilla/show_bug.cgi?id=2199

Re: [gentoo-user] CSV or mysql table as spreadsheet-like web page

[Mickaël Bucas]

Hi

There seems to be open source solutions already available :
http://en.wikipedia.org/wiki/Online_spreadsheet

I don't like to reinvent the wheel, but they may be too complex for your need.

Tell us what you choose or build yourself

Mickaël

2015-03-24 16:21 GMT+01:00 hw h...@gartencenter-vaehning.de:
 Hi,

 how would you go about creating a web page from either a CSV file or a table
 in a mysql database which presents the data to a user and lets them edit
 some of the data, preferably with the ability to use formulas like you can
 in a spreadsheet to do some calculations on the fly?

 Once editing the data is finished, it should all be saved to a table in a
 database or as a CSV file.

 Is there some php script or the like which can do this or get me started?

Re: [gentoo-user] Nvidia and Radeon video cards in the same box.

[Mickaël Bucas]

2015-02-05 17:32 GMT+01:00 Linux linux...@204eastsouth.com:
...
 [  1066.959] (II) [KMS] drm report modesetting isn't supported.
...
 [  1066.961] (EE) Screen 0 deleted because of no matching config section.
 [  1066.961] (II) UnloadModule: radeon
...
 [  1067.703] (II) AIGLX: Screen 0 is not DRI2 capable
 [  1067.703] (EE) AIGLX: reverting to software rendering
 [  1067.732] (EE) AIGLX error: dlopen of /usr/lib64/dri/swrast_dri.so failed
 (/usr/lib64/dri/swrast_dri.so: undefined symbol: _glapi_tls_Dispatch)
 [  1067.732] (EE) GLX: could not load software renderer

It seems your kernel doesn't support DRI, nor modesetting.
I would suggest checking your kernel configuration for the options
needed for radeon driver.

I don't remember how I did but there is the Gentoo Wiki :
http://wiki.gentoo.org/wiki/Radeon

Mickaël Bucas

Re: [gentoo-user] Nvidia and Radeon video cards in the same box.

[Mickaël Bucas]

2015-02-05 4:58 GMT+01:00 Linux linux...@204eastsouth.com:

 First question should be, can this be done?

 I had 2 Nvidia videos cards, running 2 monitors, with X spanning both 
 monitors.

 One card died and I replaced it with a Radeon video card.  Can I still span X 
 across the two monitors?  Does the make/model of the cards matter at all?

 If the answer is yes, then my next step is figuring out why I can get either 
 monitor to work alone, but can't get both of them to work together.

 More information will be provided once I know this is possible.

 Thanks much - Skippy

Hi Skippy

I'm currently running a system with two monitors : one attached to the
system board included radeon chip, and one attached to an nvidia card.
I've compiled the kernel with the radeon driver, and the module from
nvidia is loaded at start time.

I've tested it with a single session spanning the two monitors and it worked.
I'm using it in a multi-seat setting, with two keyboards and two mice.
I don't recommend it because it's hard to manage sound, and even
harder for OpenGL (the Gentoo wiki says it's impossible).

With two cards, you can't rely on X11 automatic configuration, so you
have to write your own xorg.conf, with PCI adresses and relative
positions of the screens :

Here is the configuration I found for single session on two screens :
Section Device
Identifier Radeon
Driver radeon
BusID  PCI:1:5:0
EndSection

Section Monitor
Identifier LG17p
VendorName LG
EndSection

Section Screen
Identifier Screen1
Device Radeon
MonitorLG17p
EndSection

Section Device
Identifier Nvidia
Driver nvidia
BusID  PCI:2:0:0
EndSection

Section Monitor
Identifier Iiyama24p
VendorName Iiyama
EndSection

Section Screen
Identifier Screen0
Device Nvidia
MonitorIiyama24p
EndSection

Section ServerLayout
Identifier default
Screen  0  Screen0 0 0
Screen  1  Screen1 RightOf Screen0
EndSection

To analyze problems, your best source is the xorg log file in your
home directory.

Mickaël Bucas

Re: [gentoo-user] lsof on fail2ban

[Mickaël Bucas]

Looking at the code, Fail2ban uses Inotify to know when a file has changed,
and only at that point it's open and read.
Inotify watches don't appear in open files.

Mickaël


2015-01-06 1:53 GMT+01:00 Adam Carter adamcart...@gmail.com:

 AFAIK fail2ban tails log files to find login failures, but when i try lsof
 its not reading daemon.log/auth.log/whatever for sshd's login failure
 messages.

 # ps -ef | grep fail2
 root   518 1  0 Jan01 ?00:05:22 /usr/bin/python3.4
 /usr/lib64/python-exec/python3.4/fail2ban-server -s
 /run/fail2ban/fail2ban.sock -p /run/fail2ban/fail2ban.pid -x -b
 root 21407 21250  0 11:45 pts/100:00:00 grep --colour=auto fail2
 # lsof -p 518 | grep var
 fail2ban- 518 root5w   REG  9,126  107   263885
 /var/log/fail2ban.log
 fail2ban- 518 root6u   REG  9,12616384  1180229
 /var/lib/fail2ban/fail2ban.sqlite3
 #

 What am I missing?

Re: [gentoo-user] OT: Backing up emails

[Mickaël Bucas]

2011/1/20 Matt Harrison iwasinnamuk...@genestate.com

 Hi guys,

 I know this isnt really a gentoo question but you always seem to come with
 an amazing
 answer in the end ;)

 I've got a friend who uses dreamhost for mail, and he's trying to move away
 from it to
 something a little more customisable.

 The problem is that he has around 2.5Gb of mail in various IMAP folders on
 this
 server. I'm trying to find some way of downloading all this mail but
 keeping in its
 current folder structure.

 I was hoping fetchmail could do it but it doesn't appear (after a quick
 read) to
 support arbitrary folders. I am almost thinking of contacting dreamhost
 support and
 asking if they can tar up the maildirs and give me access to them via FTP
 or
 something.

 I'm hoping someone here has an idea for something that will let me make a
 copy of all
 his mail, in the current folder structure and replicate it onto another
 machine.

 Grateful for any help,

 Matt Harrison


I use OfflineIMAP to backup my gmail account using IMAP.
https://github.com/nicolas33/offlineimap

It's easy to configure and to use.
Folders are preserved as it keeps mail in maildir structure.

Mickaël Bucas

Re: [gentoo-user] How can I move system to new disk?

[Mickaël Bucas]

2010/1/15 Jarry mr.ja...@gmail.com:
 Hi, I'm facing this problem:

 I want to exchange hard-drive in my computer for other, bigger
 one. I do not want to add new hard-drive somewhere on mount-point
 permanently, I just want to copy everything from the old drive
 to the new one and then get rid of the old one. And of course,
 I'd like to use my computer as before. What is the best (maybe
 I should ask for safest) way to acomplish this?

 First I thought about cp -a. But I'm not sure which directories
 I should skip (/proc, maybe some other like /dev?). And I do not
 know how cp handles links (if I first copy link and later target,
 where is the link pointing? to the original file or its copy?).

 Maybe dump/restore is better solution? Or something else?

 Jarry


I've done it twice with the following method :
http://www.faqs.org/docs/Linux-mini/Hard-Disk-Upgrade.html

There are a few things to change : no more Lilo, ext2, IDE, diskettes these days
But the check list is complete so you won't forget important points.

Mickaël Bucas

Re: [gentoo-user] What magic does portage use?

[Mickaël Bucas]

2009/12/11 Alan McKinnon alan.mckin...@gmail.com:
 On Friday 11 December 2009 13:02:36 Helmut Jarausch wrote:
 Many thanks Alan,

 so I conclude that rebooting IS necessary to get the new libraries used,
 isn't it?

 No, not at all, you conclude wrongly.

 Unix works the way it does precisely so you *don't* require a reboot to use
 new libraries. They are already there and fully installed and fully
 operational. You just have to start using them - this may require restarting
 the relevant app that uses them and perhaps ldconfig.


To find out which files have been replaced, you can use the following command :
lsof | grep DEL
This will give you all files that have been deleted since they have
been loaded by the process.
From the process name, you can deduce the service and restart it.
I've never needed a reboot for this kind of problem.
You may have to switch to run level 1 to restart some important
services like udev.

 Windows is the brain-dead johnnie-come-lately here that requires reboots. But
 then again, Windows requires a reboot when it detects the pointer has moved so
 that isn't surprising

 On the other hand running applications should continue to run, which is
 not always the case, e.g. recently using cvs as non-root user just
 hanged. Rebooting the system solved it (since I update my system nearly
 each day).

 It was probably trying to use different versions of two matched libs. You
 should not have needed a reboot to fix that.


 --
 alan dot mckinnon at gmail dot com




Mickaël Bucas

Re: [gentoo-user] Merge svnroots

[Mickaël Bucas]

2009/9/7 Song Zhiwei son...@gmail.com

 Hi all,
 Is there any way to merge a svnroot to another one? There are independent
 before.

 Regards,
 Zhiwei

Hi

I've done this using the procedure describe in SVN manual
http://svnbook.red-bean.com/en/1.1/ch05s03.html#svn-ch-5-sect-3.5

The steps are :
 - Dump all your current svn roots with svnadmin dump myrepos  dumpfile
(One file for each root)
 - Create your new svn root
 - Load your dumps into the new root svnadmin load newrepos  dumpfile

Mickaël Bucas

Re: [gentoo-user] Regenerating package db

[Mickaël Bucas]

2009/8/27 Alan McKinnon alan.mckin...@gmail.com

 On Thursday 27 August 2009 11:07:24 Xavier Parizet wrote:
  Hi everyone,
 
  I had some problem with /var/db/pkg : it's empty. So, how can i fix this
  without rebuilding world and system from scratch ?

 You need to rebuild world from scratch :-)

 Portage has now no way of knowing what you had. If emerge world does not
 work
 for some reason, you will have to:

 1. download a recent suitable stage tarball
 2. Back up everything that tarball might want to overwrite
 3. Unpack that tarball to /
 4. Put critical files back that the stage overwrote
 5. emerge world

 You might also want to do all this in a chroot off a LiveCD

 --
 alan dot mckinnon at gmail dot com


You may find some useful informations in /var/log/emerge.log : all lines
containing ::: show a succesful emerge.

If you have kept the contents of this file since you installed your system,
you could then get the list of all packages you have installed. It would be
a start that you could cleanup afterwards. However, this will include system
packages and dependencies. And if you have uninstalled many packages (as
with any update), they would also still be present in that list.

Mickaël

Re: [gentoo-user] Finding orphaned libs

[Mickaël Bucas]

2009/6/9 Joerg Schilling joerg.schill...@fokus.fraunhofer.de:
 Mickaël Bucas mbu...@gmail.com wrote:


  You may avoid the problem with find . -exec prog args {} +

 The right way to handle any size for the list returned by find, is
 by using xargs :

 find -H /usr/lib /lib -type f | xargs -d'\n' qfile --orphans

 No, this is definitely wrong: the right way to handle this is execplus
 (since 19 years).

 Jörg

I think I was too assertive.
I prefer xargs, and it's still handy for more elaborate command lines

Mickaël Bucas

Re: [gentoo-user] Finding orphaned libs

[Mickaël Bucas]

2009/6/9 Joerg Schilling joerg.schill...@fokus.fraunhofer.de

 Volker Armin Hemmann volkerar...@googlemail.com wrote:

  On Montag 08 Juni 2009, Neil Bothwick wrote:
   qfile --orphans $(find -H /usr/lib /lib -type f)
 
  qfile --orphans $(find -H /usr/lib /lib -type f)
  zsh: argument list too long: qfile

 This is not a zsh limitation but a Linux limitation.
 You may have longer arg lists if you use Solaris ;-)

 You may avoid the problem with find . -exec prog args {} +

 Jörg

 --
  EMail:jo...@schily.isdn.cs.tu-berlin.de (home) Jörg Schilling D-13353 Berlin
       j...@cs.tu-berlin.de                (uni)
       joerg.schill...@fokus.fraunhofer.de (work) Blog: 
 http://schily.blogspot.com/
  URL:  http://cdrecord.berlios.de/private/ ftp://ftp.berlios.de/pub/schily


The right way to handle any size for the list returned by find, is
by using xargs :

find -H /usr/lib /lib -type f | xargs -d'\n' qfile --orphans

xargs fills the command line up to the maximum size, and then create
another command line.

Mickaël Bucas

Re: [gentoo-user] Openttd

[Mickaël Bucas]

2008/10/21, András Csányi [EMAIL PROTECTED]:
 2008/10/21 Arttu V. [EMAIL PROTECTED]:
 On 10/21/08, András Csányi [EMAIL PROTECTED] wrote:
 Few weeks ago I reinstalled my desktop and now I want to play with
 openttd but I wonder this package is masked.
 I saw the version of this package int the portage is 0.5.3 but on the
 openttd's website the latest stable version is 0.6.3.
 So I want to ask: what is the problem whit this package?

 Hi,

 Did you read these bugs?

 http://bugs.gentoo.org/show_bug.cgi?id=215776
 http://bugs.gentoo.org/show_bug.cgi?id=233929

 They answer at least partially to many of your questions. Old one is
 blocked due to security reasons, new one has been under work and a
 version is available from that bug, they took extra time since openttd
 developers use some funky multi-platform configuration stuff of their
 own making, etc.

 Maybe you could settle for dosbox with the old TTD MS-DOS version
 until they get everything sorted out with openttd's new version? :)

 No, I didn't read. Thanks your links.
 I will be patience :)

 --
 - -
 --  Csanyi Andras  -- http://sayusi.hu -- Sayusi Ando
 --  Bízzál Istenben és tartsd szárazon a puskaport!.-- Cromwell


Hi,

From bug 233929, problem is solved in version 0.6.2.

To install version 0.6.3 I use roslin overlay, and it works for me.

Mickaël