| AlexLikeRock | hi | 00:38 |
|---|---|---|
| AlexLikeRock | help | 00:38 |
| AlexLikeRock | https://paste.debian.net/1332106/ | 00:38 |
| drizzt | re-run with "LANG=C" at begginning of command please :) | 00:43 |
| AlexLikeRock | thanks | 00:44 |
| AlexLikeRock | total agree | 00:44 |
| AlexLikeRock | https://paste.debian.net/1332108/ | 00:46 |
| AlexLikeRock | well, | 00:46 |
| AlexLikeRock | not change the lang to C | 00:47 |
| AlexLikeRock | https://paste.mozilla.org/BRpD8w41 | 00:51 |
| AlexLikeRock | https://paste.debian.net/1332109/ | 00:56 |
| AlexLikeRock | drizzt, | 00:56 |
| drizzt | AlexLikeRock: what is the command you're running ? | 00:56 |
| AlexLikeRock | aptitude install locales | 00:57 |
| AlexLikeRock | 00:57 | |
| Alverstone | apt reinstall libc-bin | 00:59 |
| Alverstone | apt-file find /usr/bin/locale | 01:00 |
| Alverstone | must be in libc-bin | 01:00 |
| Alverstone | it's not healthy not to have libc-bin | 01:00 |
| drizzt | try : | 01:00 |
| drizzt | LANG=C apt-get install locales | 01:00 |
| drizzt | to get it in english (or at least it should) | 01:01 |
| rrq | AlexLikeRock: is the back story here that you have an amd64 and want to make it multiarch? | 01:02 |
| AlexLikeRock | ok, | 01:08 |
| AlexLikeRock | E: Unmet dependencies. Try 'apt --fix-broken install' without packages (or specify a solution). | 01:08 |
| AlexLikeRock | to install libc-bin | 01:08 |
| drizzt | AlexLikeRock: that's a good starting point but check that your arch is still x86_64 (if that's what it should be) before | 01:09 |
| drizzt | dpkg --print-architecture | 01:10 |
| drizzt | (if you tried to change arch and made a mistake) | 01:10 |
| AlexLikeRock | i come , from 32 bit , and change to 64 bit | 01:14 |
| AlexLikeRock | amd64 | 01:15 |
| AlexLikeRock | dpkg --print-architecture | 01:15 |
| AlexLikeRock | amd64 | 01:15 |
| drizzt | well ... I don't know if this is possible | 01:15 |
| AlexLikeRock | yes ! , its possible but HARD | 01:16 |
| AlexLikeRock | yes, i make alot mistake | 01:16 |
| drizzt | maybe get back to 32bits, fix everything, run dpkg --add-architecture amd64, install all libs you have in :amd64 version, end the only start moving dpkg main arch to amd64, and install amd64 binaries | 01:18 |
| drizzt | but honnestly that's not something I would have tried | 01:19 |
| AlexLikeRock | https://paste.mozilla.org/JddtiXgo | 01:21 |
| AlexLikeRock | dpkg --print-foreign-architectures | 01:23 |
| AlexLikeRock | i386 | 01:23 |
| drizzt | sounds goo | 01:23 |
| drizzt | good | 01:23 |
| AlexLikeRock | its multi ARCH | 01:23 |
| AlexLikeRock | " install amd64 binaries" | 01:24 |
| AlexLikeRock | whish one ? | 01:25 |
| AlexLikeRock | libc-bin ??? | 01:25 |
| drizzt | s/end the/and then/ | 01:26 |
| drizzt | libc-bin has very little binaries in it, I mean all the other packages providing binaries | 01:27 |
| AlexLikeRock | error : whit locales ! | 01:28 |
| drizzt | all those which do not have "all" as architecture and are not lib* | 01:28 |
| AlexLikeRock | https://paste.debian.net/1332110/ | 01:29 |
| drizzt | locales is more or less part of libc6 (the locales package has same source as libc6) | 01:29 |
| AlexLikeRock | another error ! : The root PATH should typically include /usr/local/sbin, /usr/sbin and /sbin. | 01:30 |
| drizzt | AlexLikeRock: is your kernel able to run 64 bits executables ? | 01:30 |
| AlexLikeRock | It's something about path migration, | 01:31 |
| AlexLikeRock | /sbin to /usr/sbin | 01:31 |
| drizzt | AlexLikeRock: PATH=/usr/sbin:/sbin:/usr/bin:/bin | 01:31 |
| AlexLikeRock | root@DELL:~# PATH=/usr/sbin:/sbin:/usr/bin:/bin | 01:32 |
| AlexLikeRock | root@DELL:~# echo $PATH | 01:32 |
| AlexLikeRock | ---> /usr/sbin:/sbin:/usr/bin:/bin | 01:32 |
| AlexLikeRock | its ok ? | 01:32 |
| drizzt | AlexLikeRock: migration of /sbin to /usr/sbin is done by usrmerge package (though the interest of doing so is really questionnable) | 01:32 |
| AlexLikeRock | yes | 01:33 |
| drizzt | AlexLikeRock: looks right (remember that the PATH variable is local to a shell) | 01:33 |
| drizzt | AlexLikeRock: yes for the kernel being able to run x86_64 binaries ? | 01:34 |
| AlexLikeRock | i dont remember | 01:35 |
| AlexLikeRock | welll, i make CHROOOT , at my old devuan | 01:36 |
| AlexLikeRock | about kernel , can read here : https://paste.mozilla.org/JddtiXgo | 01:38 |
| AlexLikeRock | uname -a | 01:39 |
| AlexLikeRock | Linux DELL 6.1.0-25-amd64 #1 SMP PREEMPT_DYNAMIC Debian 6.1.106-3 (2024-08-26) x86_64 GNU/Linux | 01:39 |
| drizzt | well, should be :) | 01:39 |
| AlexLikeRock | yes | 01:39 |
| AlexLikeRock | the problem its the path ! | 01:40 |
| drizzt | can you run ldconfig with full path ? | 01:41 |
| drizzt | /usr/sbin/ldconfig | 01:41 |
| AlexLikeRock | ok | 01:41 |
| AlexLikeRock | how ? | 01:41 |
| AlexLikeRock | sorry | 01:41 |
| AlexLikeRock | exec .... | 01:41 |
| rrq | AlexLikeRock: did you use "su" without "-l" now again? | 01:42 |
| drizzt | yes, simply enter the command "/usr/sbin/ldconfig" | 01:42 |
| drizzt | +1 for "su" : use "su -" | 01:42 |
| AlexLikeRock | ok | 01:43 |
| AlexLikeRock | bash: /usr/sbin/ldconfig: No such file or directory | 01:44 |
| AlexLikeRock | as "su" | 01:44 |
| rrq | never, never, ever, ever use "su" without "-l" .. easy thing to remember | 01:45 |
| fsmithred | are you running excalibur? in daedalus it's /sbin/ldconfig | 01:45 |
| fsmithred | is there a way to do 'su -l' without changing directory? I'm usually already where I want to be when I become root. | 01:47 |
| AlexLikeRock_ | at , NEW_devuan : /sbin/ldconfig | 01:49 |
| fsmithred | what does new mean? | 01:49 |
| AlexLikeRock_ | I HAVE 2 DEVUAN at my laptop | 01:50 |
| drizzt | AlexLikeRock_: then can you run "/sbin/ldconfig" | 01:50 |
| AlexLikeRock_ | OLD-DEVUAN (broken down) : not found! at "/sbin/ldconfig" and "/usr/sbin/ldconfig" | 01:51 |
| rrq | so which of those chroot filesystems do you want to clean up? | 01:53 |
| AlexLikeRock_ | working .... | 01:54 |
| AlexLikeRock_ | /sbin/ldconfig: /usr/lib/x86_64-linux-gnu/libnorm.so.1 is not a symbolic link | 01:55 |
| AlexLikeRock_ | LANG=C apt-get install locales | 01:59 |
| AlexLikeRock_ | downloading .... | 02:00 |
| AlexLikeRock_ | Can't exec "locale": No existe el fichero o el directorio at /usr/share/perl5/Debconf/Encoding.pm line 16. | 02:01 |
| AlexLikeRock_ | Use of uninitialized value $Debconf::Encoding::charmap in scalar chomp at /usr/share/perl5/Debconf/Encoding.pm line 17. | 02:01 |
| AlexLikeRock_ | debconf: no se pudo inicializar la interfaz: Dialog | 02:01 |
| AlexLikeRock_ | debconf: (No hay ningún programa tipo dialog instalado, así que no se puede usar la interfaz basada en «dialog». at /usr/share/perl5/Debconf/FrontEnd/Dialog.pm line 78, <> line 781.) | 02:01 |
| AlexLikeRock_ | debconf: probando ahora la interfaz: Readline | 02:01 |
| AlexLikeRock_ | still working.... | 02:01 |
| AlexLikeRock_ | 12% | 02:04 |
| AlexLikeRock_ | 19% | 02:05 |
| AlexLikeRock_ | Cool! | 02:05 |
| AlexLikeRock_ | question : | 02:07 |
| AlexLikeRock_ | is remove 23 bit-ARCH | 02:08 |
| AlexLikeRock_ | i will lost ZNES (32bit) ? | 02:08 |
| AlexLikeRock_ | or just libs of DISTRO like LIBC , matedesktop, etc ? | 02:09 |
| AlexLikeRock_ | ZNES = my favorite videogame program | 02:10 |
| AlexLikeRock_ | my plan its remove 32 bit | 02:12 |
| AlexLikeRock_ | what its "the interface: Readline" ? | 02:16 |
| rrq | man 7 debconf | 02:21 |
| AlexLikeRock_ | No manual entry for debconf in section 7 | 02:34 |
| AlexLikeRock_ | XD | 02:34 |
| fsmithred | it's in the debconf-doc package | 02:37 |
| al1r4d | rwp :) would u drop the script? | 04:30 |
| rwp | al1r4d, What script? | 04:41 |
| rwp | We were talking about chmod and mode permission bits of /tmp? | 04:41 |
| debdog | how you hacked archive.org? hrrhrr | 04:42 |
| rwp | archive.org was breached? Sigh. I guess I must change my dog's name now. | 04:53 |
| al1r4d | > I have scripted the upgrade of my Unstable systems to run daily and to email me the result of it. I look at the email and if it worked then I just note which packages upgraded. If it failed then I go look to see why and then submit a bug report. rwp | 05:01 |
| rwp | For Sid Unstable I start by downloading and emailing me what is pending to happen. https://paste.debian.net/plain/1332120 | 05:03 |
| rwp | I get the email daily which is a good reminder and nag script to me to go and run the through the upgrades. For Unstable I run the upgrades manually after the above. | 05:04 |
| al1r4d | cool ladz, rwp | 05:07 |
| rwp | For my Stable systems I run this "apt-auto" script, which works for me but might eat your system. https://paste.debian.net/plain/1332121 | 05:08 |
| rwp | My apt-auto script I run every day from /etc/cron.daily and automatically installs security upgrades. Usually when I hear people are in a panic about some security zero day vulnerability I find that the system has already installed the patched package from the security repository. | 05:10 |
| rwp | That's not suitable for Unstable because Unstable is so often broken that one needs to give it personal attention at the time. | 05:11 |
| rwp | Speaking of Unstable... iputils-ping upgrade breaks ping. For whatever reason it is missing the iputils-ping.postinst script which sets setcap cap_net_raw+ep /usr/bin/ping or if setcap is not present chmod u+s /bin/ping and without that ping is not functional for non-root. | 09:32 |
| rwp | It's truly a shame that no one tests anything before pushing it out to the world anymore. | 09:33 |
| rwp | Oh it's worse! It's more systemd encroachment. https://bugs.debian.org/1008281 | 09:36 |
| WorldsCollide | Is extrepo safe to use in Devuan? | 10:29 |
| rwp | What is "extrepo"? | 10:31 |
| sixwheeledbeast | I believe the debian release names are written in the yaml files. | 10:31 |
| WorldsCollide | Heard about it from Librewolf's Debian instructions: https://librewolf.net/installation/debian/ | 10:33 |
| rwp | Thanks for that clarification. | 10:35 |
| rwp | Well that is a 3rd party repository. So who knows? Some 3rd parties are good. Some are not good. There is no way for Devuan to know. | 10:36 |
| helios21 | An alternative would be to install Librewolf as Flatpak. Works here. | 10:36 |
| WorldsCollide | You can install extrepo from the main repository. | 10:37 |
| helios21 | It is a bit more intense on resources to run it as a Flatpak, but also better shielded from the main system and there is no chance it could mess around with official Devuan repositories. | 10:37 |
| rwp | WorldsCollide, Since that documentation only mentions Debian the most likely problem will be that it will depend upon systemd which is explicitly NOT in Devuan. | 10:38 |
| helios21 | Not sure whether Librewolf packages would depend on an init system. | 10:38 |
| helios21 | For extrepo to work you need to install the extrepo package. So it all depends on the quality of the packages in an external repos and how well they go together with official repos. | 10:39 |
| rwp | WorldsCollide, Oh! It looks like I had a typo when I looked to see if the package was in the Devuan repository. You said it was there, I looked again, and yes it is there. | 10:39 |
| WorldsCollide | Yeah. | 10:39 |
| helios21 | I have used packaged from external Debian Multimedia repo quite a while but got totally messed up dependencies on upgrades, especially dist-upgrades often enough | 10:40 |
| rwp | Even so those are still including packages from 3rd parties. Even if the repositories are relatively curated. | 10:40 |
| WorldsCollide | If you run extrepo search, it will list all the repositories you can use. | 10:40 |
| helios21 | However it contained a lot of library packages. Not sure how many library packages the Librewolf repo would fork, if any. | 10:40 |
| WorldsCollide | Like extrepo enable winehq (for wine upstream). | 10:40 |
| helios21 | Bottom line is simple: If you use external repo you are responsible for the results :) | 10:41 |
| helios21 | So do you own risk assessment and decide. | 10:41 |
| WorldsCollide | Yea. It's the only way I could get Librewolf. | 10:41 |
| rwp | A useful command is "apt-get install apt-show-versions" which will list out all installed packages and the repository they are installed from. Useful when packages have been installed from 3rd party sites. | 10:42 |
| rwp | I often look at | 10:42 |
| rwp | I often look at "apt-show-versions | grep -v uptodate" and "apt-show-versions | grep -v /daedalus" (sorry about the accidental Enter there. | 10:43 |
| helios21 | WorldsCollide: no, again. It is not. | 10:43 |
| rwp | That will then print out any package that is from another repository. And also any package not marked uptodate. | 10:43 |
| helios21 | I have installed it as a Flatpak just fine. | 10:43 |
| helios21 | So the external repo is not the only way. Flatpak uses their own runtimes, so it is completely independent from official package management. That causes it to use a bit more resources, but at least on my laptop it is barely noticable. | 10:44 |
| helios21 | https://flathub.org/apps/io.gitlab.librewolf-community | 10:44 |
| WorldsCollide | I've never used any Flatpaks/Snaps/AppImages. | 10:45 |
| sixwheeledbeast | i think there are two issues/questions here. Can extrepo be used with Devuan? Is package foo from extrepo fine with Devuan? | 10:45 |
| helios21 | Of course if you do not like to use Flathub, then yes… the external repo it would be. | 10:45 |
| WorldsCollide | Just apt or source. | 10:45 |
| helios21 | Well the decision is whether to take the risk that installing LibreWolf via external package repository or use the more resource intensive container based approach Flatpak. | 10:46 |
| helios21 | I do not recommend snaps. | 10:46 |
| helios21 | Here are AppImage/Flatpak options documented for LibreWolf: https://librewolf.net/installation/linux/ | 10:47 |
| sixwheeledbeast | snaps are in banned due to dependencies anyway IIRC | 10:47 |
| helios21 | It may be just fine to use the extrepo. Depending on how the packages are made. | 10:47 |
| helios21 | Flatpak has probably disadvantage according to that documentation, that I just learned about: "Flatpak apps run sandboxed from the system via bubblewrap, which adds a layer of protection. But this prevents the browser from using its usual sandbox for process isolation." | 10:48 |
| helios21 | So seems isolation between system and LibreWolf is better when installed as Flatpak, but isolation between different tabs is not. | 10:48 |
| helios21 | This is mitigated a bit by "Processes are still isolated through nested seccomp filters." | 10:49 |
| helios21 | I bet that sums it up. If you decide to install by extrepo, I'd be interesting in reading how it went, WorldsCollide | 10:49 |
| freem | when it comes to web, it's better to consider that isolation is a myth, really | 14:51 |
| freem | if you do not block javascript, then you will be running code from strangers in any case | 14:52 |
| freem | and even if you do not, a simple image in your cache, even the side of a pixel, can allow to track you | 14:52 |
| freem | web is rotten, is all | 14:52 |
| freem | if web was alive once, then it now got a zombie spider, rebuilding a rotted mesh which smells foul but still looks nice enough | 14:56 |
| onefang | Doesn't help that Firefox-ESR will check each page you load with Google, to see if it's "safe". | 15:14 |
| onefang | Back on topic though, I'm trying to write a simple sys-v init script. It's just not getting called. though the proper links are getting created by sysv-rc-conf.. Is there something new I'm missing that needs to be done? | 15:16 |
| onefang | It's in /etc/init.d. It has the proper LSB pre-amble. sysv-rc-conf recognises it, lets me turn it on and off, which then creates the links. | 15:18 |
| Alverstone | why do I have 50+ old log files with svlogd? | 15:19 |
| Alverstone | it says it only leaves 10, in manpage | 15:19 |
| Alverstone | default is 10 | 15:19 |
| Alverstone | not 50+ | 15:19 |
| Alverstone | is it a bug | 15:19 |
| fsmithred | Alverstone, check logrotate settings | 15:35 |
| gnarface | onefang: does the script have execute permissions? does the name of the script end in .sh or not? | 15:48 |
| gnarface | is it perhaps a copy of another script that checks a file in /etc/default/ for an enable flag? | 15:49 |
| gnarface | Alverstone_: (make sure logrotate is even installed, i seem to remember it going missing during an upgrade at some point...) | 15:50 |
| onefang | I figured it out. Running sysv-rc-conf for a NEW init script isn't enough, you need to run update-rc.d on it first. | 15:52 |
| onefang | Only some of the init scripts end in .sh, so that doesn't matter. | 15:54 |
| onefang | Plus I want to write mine in Lua, so .sh is silly. lol | 15:54 |
| onefang | You can write sys v init "scripts" in any language, so long as the proper LSB pre-amble is in place. I've written them in C in the past. | 15:59 |
| gnarface | yea, but i remember some gotcha at some point where the ones with .sh in the name were being treated differently... i forget the specfics | 16:08 |
| gnarface | (might have only been relevant to an older release though) | 16:08 |
| Alverstone_ | gnarface, svlogd depends on logrotate!? | 16:19 |
| fsmithred | logrotate is one of the packages that needs to be excluded in a debootstrap build of excalibur or ceres right now, along with cron-daemon-common. But they can be added afteward. | 16:22 |
| fsmithred | looks like logrotate is a Recommends, not a Depends. | 16:24 |
| gnarface | Alverstone: i don't know for sure but i assume so | 16:25 |
| Alverstone | but logrotate only rotates logs which are listed in configs | 16:25 |
| Alverstone | and those are not the ones made by svlogd | 16:26 |
| Alverstone | and logrotate doesn't even work in the same manner | 16:26 |
| Alverstone | svlogd uses special naming for rotated logs | 16:26 |
| Alverstone | anyway, installing it didn't help | 16:32 |
| gnarface | well you might have to configure it to look for your logs | 16:33 |
| gnarface | obviously you wouldn't want anything double rotated though, so i guess check the svlogd docs | 16:34 |
| onefang | gnarface: I suspect any gotcha with not being .sh init scripts might have been some long ago bug. My new .lua init script is now working fine. I just have to actually add the guts of it now. | 16:40 |
| Alverstone | okay so I added n5/n10 to svlogd config files and it works | 16:43 |
| Alverstone | guess the real default is 0 for some reason | 16:44 |
| onefang | Actually the only reason I put the .lua extension on is so my text editor knows it's dealing with a Lua script and syntax highlights properly. | 16:46 |
| rwp | JFTR but Debian Policy documents NOT using suffixes such as .sh or .pl or .py and the others specifically to allow other implementations. You will know someone was not thinking when they froze the program name to be foo.sh and looking inside it is a python script! | 19:28 |
| rwp | onefang, update-rc.d is what updates the symlinks stitching it into the directory where it is called. I assumed that sysv-rc-conf would do this, no? How does sysv-rc-conf work then? | 19:30 |
| fsmithred | sysv-rc-conf [ --level levels ] service <on|off> | 19:36 |
| fsmithred | rwp, it looks like it's possible and only slightly simpler than making the simlinks manually. | 19:36 |
| rwp | I have not used sysv-rc-conf but it does look like it should be okay and do everything. | 19:38 |
| fsmithred | for turning stuff on and off it's brain-dead simple | 19:39 |
| rwp | The update-rc.d was originally created for use in package postinst scripts. So that a package installing would call that as a hook to Do The Right Thing for whatever was installed. | 19:39 |
| fsmithred | arrows, space bar to make a check and q to exit | 19:39 |
| fsmithred | thank you whoever invented that | 19:39 |
| rwp | Which is why the arguments to update-rc.d are somewhat unfriendly. It's designed to be called from the postinst script and it's easy there to copy-paste something that works and never call it from the command line. | 19:40 |
| rwp | I installed sysv-rc-conf and looking at the man page for it, it does look like either sysv-rc-conf or update-rc.d do the same thing. Running either should be sufficient. | 19:42 |
| rwp | Also sysv-rc-conf --list appears almost identical to "chkconfig --list" output. chkconfig is a tool that was dropped from the repos a couple of releases ago. But it is just a shell script. So I drag it forward in /usr/local/bin for me. | 19:44 |
| rwp | Ha! "sysv-rc-conf can also be used at the command line when the desired changes to the symlinks are already known. The syntax is borrowed from chkconfig(8), although it does not follow it exactly." so that's why it is so close in action to chkconfig. | 19:45 |
| fsmithred | hey, I have a mail question | 19:45 |
| fsmithred | I fixed /etc/mailname in a couple of recent iso builds, and I also ran dpkg-reconfigure exim4-config. | 19:46 |
| fsmithred | Where is asks who should get system mail, I left it blank. If I put in the current user's name (user) and they change their username when they install, will exim send them the mail? | 19:47 |
| rwp | fsmithred, I agree with what you said about sysv-rc-conf in that it works brain dead simple. I tested using it to disable cups-browsed (recent security vulnerability) and it worked swimmingly. | 19:47 |
| fsmithred | yeah, I've been using it for years to turn stuff off that I don't use often. | 19:48 |
| rwp | It's really a fully functional replacement for the previous chkconfig (which I have been continuing to use regardless of it not being in the repos) but sysv-rc-conf is in the repos and easily available to people. I AM CONVERTING! At least in my instructions to other people. :-) | 19:49 |
| fsmithred | cool | 19:49 |
| fsmithred | brb, I need to switch computers | 19:51 |
| rwp | Regarding your two questions: /etc/mailname is a shared place to put the system host name used for outgoing mail headers that all of exim4 and postfix and all of the others can use to set their host name in a shared way. So that the dialog interface doesn't need to know details about each. Postfix has a Debian specific code patch to pull that file in to the config. I assume the rest are similar. The file is non-standard but it was a | 19:51 |
| rwp | way to have a dialog question work for all of the MTAs that might be installed. | 19:51 |
| fsmithred | I changed /etc/mailname to localhost | 19:52 |
| fsmithred | I left system recipient blank, and I think it goes to /var/mail/mail now. | 19:53 |
| rwp | Maybe localhost.localdomain instead? Let's come back to that question in a moment. | 19:53 |
| rwp | The second question is about the root email alias in /etc/aliases and where that should go. People don't like requiring root to read system email output. So they want to always have root mail aliased to go to a non-root account. | 19:53 |
| fsmithred | maybe. I've never voluntarily used a domain name | 19:53 |
| fsmithred | can that file have uid:gid instead of a name? | 19:54 |
| rwp | The email alias is one of those things that was simple in 1999 when people read email on the system but is now problematic when no one reads email anywhere and if they do it is at google or yahoo servers via a web browser. Connecting the dots there is a problem. | 19:54 |
| fsmithred | lol, I only use it for system mail | 19:55 |
| rwp | The /etc/aliases file cannot have a uid:gid there. It's a mapping lefthandside to righthandside with something like "root: rwp" or "root: bob@proulx.com" there. Except actually sending mail off machine to another machine requires that you read ten years of information about how to navigate other system's antispam policies. | 19:56 |
| fsmithred | lol | 19:56 |
| rwp | IF you know the local user name THEN I would map it there. "root: refracta" and have it delivered to the local refracta user, for example. "root: rwp" for me for example. Then it would go to /var/mail/rwp and I would read it with "mutt" as my rwp user account and not need root for it. That's probably the best compromise. | 19:57 |
| fsmithred | yeah, I have some scripts on a remote server that send me email, and I had to switch from sending it through google. | 19:57 |
| fsmithred | So if I set it to "user" because that's the username, and they install and change the name, the mail goes to a box named user? | 19:57 |
| rwp | I don't want to discourage anyone from running their own mail server system but it's not something that be flipped on in a configuration file option. It's an entire infrastructure build to do it. | 19:58 |
| rwp | Yes. The mail will go to /var/mail/user in that case. | 19:58 |
| fsmithred | and the newly named user will still have permission? | 19:58 |
| rwp | "and they install and change the name" I lost sync here. They are still using "user" or they are using something "fsmithred" there? | 19:59 |
| fsmithred | changed name | 19:59 |
| fsmithred | assume refractainstaller here | 19:59 |
| fsmithred | file permissions are by number, so they should still be able to read it | 20:00 |
| rwp | Then it will go to /var/mail/fsmithred instead if that is both a) a valid user account name and b) the mapping line in /etc/aliases "root: fsmithred". | 20:00 |
| fsmithred | how does that get changed from root: user? | 20:00 |
| fsmithred | I guess I need to test this | 20:00 |
| rwp | Oh now I understand your uid:gid question. Yes the underlying file is using the numbers. But the MTA mail-transport-agent passing the email to the MDA mail-delivery-agent will use the name to validate or bounce the message. MDA is like exim4 or postfix or dma or nullmailer or something and MDA is like procmail or sieve or something. | 20:01 |
| rwp | The installer asking where to deliver root email will set /etc/aliases with the map and then whenever /etc/aliases is modified the "newaliases" command is run to freeze it into the /etc/aliases.db database file for fast lookup of the mapping. | 20:02 |
| fsmithred | my installer does not ask that question | 20:02 |
| rwp | I admit I am not as familiar with your installer as I should be to answer the question. :-) | 20:03 |
| rwp | But if the /etc/aliases root alias matches an actual non-root account then everything will work. | 20:03 |
| fsmithred | rsync copy to disk, chroot to add bootloader, ask about changing passwords and name, and it goes through the user's home dir configs to change the name. | 20:03 |
| fsmithred | I don't do anything to change /etc/alias. | 20:04 |
| fsmithred | es | 20:04 |
| rwp | What would change in the user's home directory? | 20:04 |
| fsmithred | stuff in .config that refers to /home/$USER | 20:05 |
| fsmithred | and unfortunely, using that variable won't work | 20:05 |
| rwp | Hmm... ".config" hmm... What's in there that would need to change? (Me contemplates "rm -rf ~/.config" as the grumpy Luddite that I am here.) | 20:05 |
| fsmithred | so when you open a terminal or file manager or text editor it know what to use for working dir | 20:06 |
| rwp | It doesn't open a terminal or editor in the current working directory where it was invoked? | 20:06 |
| fsmithred | it remembers to open in a directory that no longer exists | 20:08 |
| fsmithred | because the name changed | 20:08 |
| fsmithred | grep --color -r "/home/$USER" ~/.config | 20:10 |
| rwp | And this is something that can't be solved by "rm -rf $USER/.config" there? | 20:10 |
| fsmithred | try that command | 20:10 |
| fsmithred | mine, not yours! | 20:10 |
| fsmithred | yes, that would solve it and wipe out all the desktop settings back to defaults for the desktop | 20:11 |
| fsmithred | which pretty much negates the point of using refracta tools. | 20:11 |
| rwp | I try your command (not mine, haha) and I get this after I filter some obvious stuff out. https://paste.debian.net/plain/1332189 | 20:13 |
| rwp | For me there is nothing there of interest. All of that could be removed. | 20:13 |
| rwp | Xfce will recreate a bunch of it the first time it runs if it is not there. Which is annoying. Even if a user removes the files they keep coming back if using a DE. | 20:14 |
| fsmithred | mine scrolls way off the screen | 20:14 |
| rwp | I had to grep -v several thigns wqhich I showed there in order to stop that from happening. But it was all google chromium related. | 20:15 |
| fsmithred | some is settings some is recent files used by different programs | 20:15 |
| rwp | I think you should seriously consider not shipping any files in $USER/.config at all and let the DE populate it when it starts. | 20:15 |
| rwp | It is exactly the same situation as when a new user is created. It has an empty home directory. Then the user logs in and starts a DE and all of those files become populated. | 20:15 |
| rwp | But it starts from an empty directory. | 20:15 |
| fsmithred | yeah, then I have to deal with /etc/xdg | 20:16 |
| rwp | Why would you need to deal with changing defaults in /etc/xdg? What defaults are you wanting to change? | 20:16 |
| fsmithred | what's in the panel mostly | 20:16 |
| fsmithred | also theme and background | 20:17 |
| rwp | If you want to change those defaults then I think you should be changing them in /etc/xdg in the system level configuration. Because what happens when a new user is created and they log in as that new user? | 20:18 |
| rwp | Are you populating /etc/skel/ too? | 20:18 |
| helios21 | I think Plasma creates the initial folders only once… but I am not sure. At least you can configure alternative locations for documents, music, videos and so on and it uses those then | 20:18 |
| helios21 | and with ~/.config as well as ~/.local/share… any app can write stuff in there. I have given up on cleaning up stuff in there. Sometimes when I noticed something from a program I uninstalled years ago I may remove it… | 20:19 |
| fsmithred | yeah you can set the default directories in /etc/xdg. I do that to get rid of the default mess. | 20:20 |
| helios21 | advantage is at least… when its in above directories it is not scattered around in some other hidden directory. My $HOME is a mess. It is has the accumulated stuff of several decades meanwhile as I always copy it over to a new machine. | 20:20 |
| fsmithred | not using /etc/skel. | 20:20 |
| rwp | Sometimes I screw up interactive settings for htop and I "rm ~/.config/htop/htoprc" as a quick way to reset back to a known starting state. | 20:20 |
| helios21 | well in Plasma you can set the directories in Systemsettings directly. | 20:20 |
| rwp | Since you are not using /etc/skel/ which is the template for creating new user accounts it means that if a new user account is created it will not have any ~/.config at all at the start. They will log in. The DE that starts will start populating everything from not being there. It won't have any of the changes you initially set up for that first user. | 20:21 |
| helios21 | rwp: haha, yeah, after I found out I can Ctrl-W write current configuration in top I did that with top as well a few times. even config for top has moved to ~/.config. It was in main home directory. And funnily enough the programs usually do not clean up. They just migrate to the new location but leave the old one intact as well. | 20:22 |
| fsmithred | yeah, I know about that problem with new user accounts | 20:22 |
| rwp | I am not sure it is a problem as such. It's just the way things work. | 20:23 |
| helios21 | that way you could have the same configuration file in three different locations but only one of them is used | 20:23 |
| helios21 | rwp: well sure… that is why I have just given up on aiming at cleaning everything up. | 20:23 |
| helios21 | If I notice something by accident I usually clean it up. Everything else resides where it is. Those files are usually not large, so why bother. | 20:24 |
| rwp | helios21, Yes. It's a problem! All of that cruft and lint remaining behind. | 20:24 |
| helios21 | Ah, you meant fsmithred's issue? | 20:24 |
| helios21 | rwp: well how to solve? I read about a program cleaning up stuff in Linux, but I generally do not trust these kinds of programs. | 20:25 |
| fsmithred | bleachbit? | 20:25 |
| helios21 | I think that was the one. Not sure how good it is with cleaning configuration files from obsoleted locations. | 20:25 |
| rwp | I haven't looked at bleachbit but I understand the motivation for it. But right they are almost always just not quite perfect. | 20:26 |
| helios21 | You'd basically need a database with thousands of locations of former configuration files. You also need to check whether they really migrated to new places… and so on. | 20:26 |
| helios21 | Its not an easy task to solve. | 20:27 |
| rwp | I am getting called away. I want to summarize that I don't think one needs to edit/hack files in a ~$USER/.config directory. I think it is okay to have the DE and tools start from an empty directory. That's okay. (That's preferred for me.) | 20:28 |
| rwp | The mapping from /etc/aliases echo "root: $USER" >> /etc/aliases ; newaliases; works if $USER is a valid user. Then mail is delivered to /var/mail/$USER okay. | 20:28 |
| rwp | The /etc/mailname I think should use "localhost.localdomain" for reasons that I would enjoy talking through in a bit when I am back with time. It depends upon /etc/hosts file coordination though. But you can do that easily too. And then it is all self-consistent. | 20:29 |
| rwp | BBIAB. | 20:30 |
| fsmithred | ok, thanks | 20:32 |
| onefang | Not sure if this was answered, but the thing that update-rc.d does that sysv-rc-conf doesn't do is to update those /etc/init.d/depends* files. | 20:33 |
| onefang | Was at the other computer in the other room fiddling with this stuff, so only just noticed now. | 20:34 |
| fsmithred | meet me in dev | 20:35 |
| AlexLikeRock | hi | 21:49 |
| Alverstone | \o | 22:15 |
| Xenguy | o/ | 22:49 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!