EHeM | Missing a "1400 packages will be upgraded" message is a known risk of "stable"; seems a bit hard since the list of packages would be kind of huge and even over a very fast link it would take a while to grab all the packages. | 02:17 |
---|---|---|
koollman | well ... if you have a local repo, and 2x10g links or something like that ... (and nvme drives ... ) :) | 02:18 |
EHeM | KatolaZ: Isn't Devuan supposed to be about choice and not needlessly breaking things? "stable"/"testing" are discouraged for legitimate reasons, but they're handy for some and hence really shouldn't be broken... | 02:18 |
EHeM | koollman: Yes, in that case it is possible. | 02:19 |
EHeM | I don't have nearly that much bandwidth, so that would be downloading for a while. | 02:19 |
golinux | EHeM: That won't work when Debian is stable and we're still in testing | 02:20 |
golinux | on the way to stable | 02:20 |
golinux | And amprolla is not set up to deal with suites. It set up for rele4asenames | 02:20 |
gnarface | look the fact is debian strongly advises against it too, because you can still break a debian system with it too | 02:21 |
golinux | to avoid that happening | 02:21 |
EHeM | "Amprolla" hasn't hit my radar. | 02:21 |
gnarface | so there's an additional risk with devuan, but that doesn't really mean it's a devuan specific issue that you can break your system if you set it to "stable" and it changes out from underneath you unexpectedly | 02:22 |
EHeM | gnarface: Legitimate ones too, but it is handy for many circumstances. | 02:22 |
gnarface | in both cases the risk is probably bigger with testing | 02:22 |
fsmithred | I still don't understand what advantage you get from using "stable" | 02:23 |
gnarface | people just want to make sure you fully understand the risks, because this is the type of thing that gets frequently forgotten by the time it causes an actual problem | 02:23 |
EHeM | Many of the issues could be heavily simplified by making some level of substitutions in sources.list possible (change 1 string on one line, rather than strings on >10 lines for each update). | 02:23 |
fsmithred | ok, that makes sense, except it won't work | 02:24 |
fsmithred | you'll have to change them to codenames before buster reaches stable or you'll have a broken system | 02:24 |
* EHeM has still yet to hear a justification of why preserving functionality is impossible. | 02:26 | |
gnarface | i heard them say it | 02:26 |
fsmithred | not gonna be easy when kernel, libc6 and some other stuff are at buster versions and all the packages devuan changes (e.g. util-linux, bsdutils) are still at ascii versions | 02:27 |
gnarface | it changes the case of "debian unexpectedly updates which one stable points to" from "might break" to "definitely breaks" | 02:27 |
fsmithred | EHeM, get a test system that's like the systems you have to maintain. Add beowulf repo but keep ascii repos in sources.list. Then dist-upgrade. Rinse and repeat periodically until buster or beowulf goes stable (whichever is first) and you'll have some advance knowledge of the problems you'll run into. | 02:33 |
* golinux needs to do that too | 02:34 | |
xrogaan | shouldn't devuan be listed in there? https://community.kde.org/Distributions | 03:20 |
EHeM | This makes it sound like the Devuan package download servers need to know what "stable" is in order to tell clients what to download (ie download "Debian/distro" instead of "Debian/stable"), I'm wanting to suggest I vaguely recall seeing mentions of functionality already existing. | 04:43 |
golinux | EHeM: Perhaps you should read the amprolla code and submit a patch to have it do exactly what you want. If not please just use the repos are they were intended to be used. | 04:49 |
EHeM | You're making it sound like the package servers were setup with "stable" setup as being a distinct distribution, rather than being a reference to whatever was the current stable release; this sounds like poor server setup. | 04:58 |
fsmithred | the servers are set up so that the packages altered by devuan come from devuan servers, and all the rest come from debian servers. Amprolla acts like a filter to keep out the bad stuff. | 05:00 |
fsmithred | right now, devuan's stable is pulling from debian's stable. | 05:01 |
fsmithred | debian's current testing will probably become stable before devuan's testing does. | 05:02 |
EHeM | fsmithred: Problem is Amprolla should be mapping requests for Devuan/stable/package => Debian/<corresponding release>/package, whereas those are being passed straight through. | 05:02 |
fsmithred | when that happens, amprolla will pull packages from debian oldstable for ascii | 05:02 |
fsmithred | sorry, I don't understand | 05:03 |
fsmithred | and I'm probably saying it wrong: amprolla will continue to pull from stretch for ascii | 05:04 |
fsmithred | the only way it would be safe to use stable/testing/unstable instead of codenames is if devuan and debian released the stable version at the same time | 05:05 |
Chanku | To explain it. Amprolla is designed to manage which packages are changed by Devuan (and thus pulled from Devuan's repos) and those that aren't (and pulled by Debian's repos). | 05:06 |
Chanku | Amprolla has no concept of stable/testing/unstable | 05:06 |
fsmithred | those names are a moving target | 05:06 |
Chanku | When you request something from 'stable' to amprolla it will check and see what 'stable' packages exist in the Devuan Repo and Debian Repo, however when Buster hits stable, Amprolla will merely pass the 'stable' request to the new stable for Debian repos | 05:07 |
EHeM | fsmithred: Sounds like it is mapping Devuan/ascii => Debian/stretch; but Devuan/stable => Debian/stable instead of Devuan/stable => Debian/stretch. | 05:07 |
Chanku | which would be 'Buster' | 05:07 |
EHeM | Chanku: What you are describing is NOT possible, if it didn't map distribution names, then clients would ask the Debian servers for Debian/ascii which does NOT exist. | 05:09 |
EHeM | Chanku: Therefore the ability to map names MUST already be present. | 05:09 |
Chanku | I'm saying Amprolla doesn't know what stable/testing/unstable are at all. It knows what ascii/jessie/stretch/buster is | 05:10 |
EHeM | Chanku: Which sounds like a configuration problem. | 05:10 |
Chanku | Have you looked at Amprolla at all? | 05:10 |
EHeM | (which should be easy to adjust) | 05:10 |
Chanku | If not you are speaking out of your ass and making assumptions | 05:10 |
fsmithred | yeah, the configuration problem is that debian stable and devuan stable may or may not be the same | 05:10 |
fsmithred | therefore it makes sense not to use those names | 05:11 |
EHeM | Chanku: Nope, but two strings not matching is already present and therefore the capability must already exist (could have been implemented badly though). | 05:11 |
Chanku | But in any case, stable/testing/unstable are moving targets that Amprolla doesn't know about, and IIRC isn't necessarily likely to be implemented due to Amprolla. | 05:11 |
Chanku | EHeM: then by all means | 05:11 |
Chanku | submit a PR for Amprolla to do this, however this is expected and known behavior. | 05:12 |
EHeM | Chanku: Which location is prefered for reportting? | 05:13 |
fsmithred | bugs.devuan.org | 05:14 |
Chanku | I think he was asking for submitting a PR for amprolla | 05:14 |
golinux | EHeM: Arguing is not a good use of your energy. Write a patch or move on. | 05:14 |
fsmithred | oh, then I guess git.devuan.org | 05:15 |
golinux | This is a do-ocracy. You have spent quite some time talking to KatolaZ so i's unlikely any of the Devuan devs have an interest in fixing YOUR problem | 05:16 |
golinux | i's > it's | 05:16 |
golinux | Continuing to whine will not make it happen. | 05:17 |
golinux | Because it's not broken | 05:18 |
EHeM | golinux: A patch for a configuration file seems overkill, but such is doable. | 05:18 |
golinux | If it has merit it will be considered. | 05:19 |
EHeM | golinux: I'm also trying to be sure to point the finger at the right location (right now this sounds like a simple configuration tweak, but you could convince me otherwise). | 05:20 |
golinux | I am the wrong one to ask. You would need to convince KatolaZ and so far that hasn't happened. | 05:23 |
golinux | Neither has he (or others) convinced you so stalemate. | 05:24 |
golinux | You're talking past each other. Code would clarify what you want done and will stand on its own merit. | 05:25 |
golinux | (Don't hold your breath) | 05:27 |
EHeM | golinux: If additional executable code (excluding shell fragments due to many configuration files being parsed by a shell and can therefore contain executable code despite not being supposed to), I would be astonished. | 05:29 |
golinux | https://git.devuan.org/devuan-infrastructure/amprolla3 | 05:32 |
golinux | Have fun! | 05:32 |
Centurion_Dan | what' | 06:11 |
Centurion_Dan | the #$%^ | 06:11 |
Centurion_Dan | why on earth won't EheM listen. He's reported an issue that we won't fix because he's proposing something dangerous. | 06:13 |
golinux | Because he doesn't get it. | 06:14 |
golinux | All he wants is his workflow to be a certain way. | 06:15 |
Centurion_Dan | EHeM: we wlll not guarantee or encourage the use of "stable" or "testing" labels, for the shear fact that it upgrades aren't always simple and there are almost always decision points where a package is deprecated or introduces a new version in the new release that doesn't have an automated upgrade path. It's as simple as that. | 06:17 |
* golinux gets popcorn | 06:17 | |
Centurion_Dan | As for how amprolla works... It only ever redirects for debian packages to the version in the equivalent suite in debian - this is because the redirect is for the the file to be downloaded from the pool. Debians list of packages, codenames, labels or other suite metadata are never seen by apt when it uses only devuan repos in it's sources.list | 06:20 |
Centurion_Dan | EHeM: some mirrors in the roundrobin are broken because they happen to use lighttpd and can't handle symlinks which is why it breaks. Personally I don't think we should allow such mirrors, but I'm not the one calling the shots on that. | 06:22 |
Centurion_Dan | (Alternatively the lighttpd installations could manually insert redirects from stage names to codenames - but again they need to be changed in sync with new releases. | 06:26 |
Centurion_Dan | ) | 06:26 |
Centurion_Dan | But this is simply beside the point. You've been told to use release names not 'stable' or 'testing' etc simply because it's bas practice which will result in loud complaints when an automated update suddenly breaks your system because the suites changed and you' | 06:30 |
Centurion_Dan | you've ended up with a failed upgrade or worse a successful upgrade that silently removed applications or broke expected behaviour both of which you can't walk back because the upgrade has gone too far. | 06:33 |
EHeM | Centurion_Dan: I (and other above) knew of the danger that "stable" could cause distribution updates unexpectedly, and need to be prepared to deal with that. For large clusters, yes that is a big danger, for one's smaller personal systems (also for test systems) the handiness is much greater than the risk of a screwed up system. | 06:35 |
EHeM | The alternative method is for APT to handle macros in sources.list (ie set my_dist=ascii, then later use ${my_dist}) so updating sources.list isn't a pain if you've got more than 10 lines in there. | 06:37 |
Centurion_Dan | Fine, you acknowledge you get to keep the pieces.... best thing to do then is pick a working mirror out of the pool that is close to you that works and stick with it rather then rely on the roundrobin. | 06:37 |
Centurion_Dan | EHeM: You might want to try the Deb822 format replacing /etc/apt/sources.list with | 06:51 |
Centurion_Dan | /etc/apt/sources.list.d/devuan.sources | 06:51 |
Centurion_Dan | and containing something along these lines: | 06:51 |
EHeM | Centurion_Dan: Too much text has gone by for me to cite the statement, but I'm pretty sure it was mentioned not to be a lighttpd issue; failing that, it is only a single mirror which is failing. | 06:51 |
EHeM | (so there is a dislike of lighttpd and it is only on a single mirror? nuke it.) | 06:52 |
Centurion_Dan | " Types: deb deb-src | 06:53 |
Centurion_Dan | URIs: http://deb.devuan.org/merged | 06:53 |
Centurion_Dan | Suites: ascii ascii-updates ascii-security | 06:53 |
Centurion_Dan | Components: main contrib non-free | 06:53 |
Centurion_Dan | " | 06:53 |
Centurion_Dan | That would allow you to only change oneline. | 06:54 |
Centurion_Dan | Then again, even with the current format, if you use vim as the editor you can change the suites with a single command: | 06:54 |
Centurion_Dan | :%s/<old suite>/<new suite>/ | 06:55 |
Centurion_Dan | EHeM: if we had enough mirrors that meant it wasn't really a loss, I'd also suggest to the mirror maintainers to drop it from the round robin - but I don't know if that's a reasonable thing to do at this stage. | 06:58 |
Centurion_Dan | But for your issue... if you want to use 'stable' or 'testing' etc as the suite selector, then pick a mirror out of the list http://pkgmaster.devuan.org/mirror_list.txt that works for you and stick to that. | 07:01 |
EHeM | Multiple suites are allowed on a single line? That is a change (perhaps it has been allowed for a long time, but some of my configuration files do trace their lineage back more than a decade). | 07:34 |
EHeM | This though leaves the issue of you have to change the same string multiple times (yes, a simple change, but still redundancy). | 07:35 |
EHeM | I tend to keep files in RCS, so repeated changes increases the history... | 07:36 |
Centurion_Dan | you could always use the same vim command but with a "g" added after the last "/" | 07:41 |
Centurion_Dan | to get the same effect. | 07:41 |
KatolaZ | EHeM: just avoid to use deb.devuan.org | 07:43 |
KatolaZ | stick to a single mirror that works for you | 07:43 |
KatolaZ | devuan uses codenames, and encourages users to use codenames | 07:43 |
KatolaZ | because this is the easiest way to avoid serious breakage | 07:44 |
KatolaZ | there is nothing else to say about that | 07:44 |
KatolaZ | and, BTW, amprolla is not *serving* anything | 07:45 |
KatolaZ | amprolla is just merging repositories | 07:45 |
Xenguy | Don't fear the Reefer: https://www.youtube.com/watch?v=PVn6b9QQZeM | 07:52 |
terra | Guys, which package is responsible for usb auto mount on pcmanfm/LXDE ? | 08:14 |
terra | I installed LXDE manually but usb drives are not recognizing by pcmanfm | 08:15 |
Xenguy | I would think that should be default behavior i.e. hotplugging of removable media | 08:16 |
Xenguy | But not running LXDE currently (tried it before tho). MATE works, FWIW | 08:16 |
KatolaZ | terra: you should fiddle with polkit-1* and elogind/consolekit | 08:18 |
KatolaZ | the reason why LXDE is not one of the deault choices is that it seems to have problems with either consolekit or elogind | 08:18 |
terra | I installed with recommendations disabled | 08:19 |
gnarface | terra: is this that install you wanted to keep below 500MB? | 08:24 |
terra | gnarface: exactly. I installed gvfs then usb mount works now. | 08:36 |
terra | that cost me about 20MB :) | 08:36 |
gnarface | ouch | 08:37 |
gnarface | well there may be other options | 08:37 |
terra | udisks2 doesn't help | 08:37 |
terra | or I'm doing something wrong | 08:38 |
terra | I use lxdm instead of lightdm | 08:38 |
gnarface | heh, i'm actually still using e17 | 08:39 |
gnarface | so it could be a difference in window managers | 08:39 |
terra | I'm 20+ years linux user. Still couldn't figure out how plkit, colsolekit, udisks, udev, gvfs mess | 08:40 |
KatolaZ | terra: you need a session manager (elogind/consolekit) + the corresponding polkit backend | 08:40 |
terra | ck-list-sessions was listing my session | 08:41 |
terra | but no usb | 08:41 |
terra | this things must handle by display manager eg: lxdm | 08:42 |
KatolaZ | terra: the reason why LXDE is not among the installable options in ascii is that the config was not reliable | 08:42 |
terra | hmm, I see. | 08:43 |
KatolaZ | I guess the problem was the combination of login manager + session manager | 08:44 |
terra | I tried Slax yesterday. Its simplicity and effectiveness impressed me. | 08:44 |
terra | If that was fluxbox, I never used fluxbox before. | 08:45 |
KatolaZ | terra: I am the worst person to ask about this stuff though | 08:45 |
KatolaZ | I am using wdm + xmonad | 08:45 |
KatolaZ | (or wdm + wmaker) | 08:45 |
KatolaZ | and I still use pmount | 08:46 |
terra | KatolaZ: AfterStep was my first WM | 08:46 |
KatolaZ | <3 | 08:46 |
KatolaZ | ;) | 08:46 |
terra | Hope GNUStep gains the popularity it deserves | 08:46 |
KatolaZ | terra: I hope it doesn't, if popularity implies the degradation I have seen in almost any "popular" DE.... | 08:47 |
terra | KatolaZ: you are propably right. Just like what happened mainstream Linux distros. | 08:48 |
terra | Thinking about all mess aroung switching to gtk3 ..then it comes gtk3. | 08:51 |
terra | *gtk4 | 08:51 |
terra | I still don't know what gtk3 brings and why wee need gtk4 | 08:53 |
amesser | terra: when ck-list-sessions showed your session and you have installed polkit backend consolekit, user mounting should work | 09:07 |
amesser | you could try mounting from commandline udisksctl mount -b /dev/sd?? | 09:08 |
amesser | if that works, problem is lxde, if it does not work problem is session/polkit | 09:08 |
terra | amesser: is this alternative for gvfs? It seems gvfs is using udisks and things get corrected by installing both | 09:12 |
amesser | i dont know much about gvfs. But the actuall mounting work is done by udisks. On KDE, the file manager asks udisks to mount a drive. The same happens when using the command line tool "udiskctl" | 09:13 |
amesser | maybe the lxde filemanager relies on gvfs | 09:14 |
terra | in short: can I do without gvfs? or can you do? can you check "ps aux|grep gvfs" ? | 09:14 |
amesser | ohh, semems gvfs is installed here also. let me get my old notebook, Using i3 there | 09:15 |
amesser | man udisksctl should be working without gvfs | 09:15 |
amesser | results for my old notebook: | 09:17 |
amesser | $ ps aux | grep gvfs | 09:18 |
amesser | $ udisksctl mount -b /dev/sdb2 | 09:18 |
amesser | Mounted /dev/sdb2 at /media/andi/devuan_2.0.0-rc.amd64_n. | 09:18 |
amesser | $ | 09:19 |
amesser | So it works with gvfs | 09:19 |
amesser | (all commands executed as user) | 09:19 |
terra | amesser: are you using pcmanfm or thunar ? | 09:19 |
terra | I mostly interedted on File Manager/ DE integration | 09:20 |
amesser | this was command line only. I can install them and test which one do you prefer | 09:20 |
amesser | One difference maybe: I'm using elogind, not consolekit | 09:20 |
terra | you could try pcmanfm and standalone i3 desktop for instance | 09:20 |
amesser | ok | 09:20 |
terra | <amesser> So it works with gvfs --> without ? | 09:21 |
amesser | Works without :-) Hmm, pcmanfm pulls in some gvfs packages | 09:22 |
terra | you must install pcmanfm without recommendations as described here: https://devuan.org/os/documentation/dev1fanboy/migrate-jessie-minimalism | 09:23 |
terra | Check out "Configure minimalism" section | 09:24 |
amesser | ah ok, i ll unistall gvfs again | 09:24 |
amesser | No mounting with pcmanfm. Maybe it relies on gvfs to do the mounting | 09:30 |
terra | amesser: thank you. you just confirmed that :) | 09:31 |
amesser | let me try thunar | 09:36 |
Unit193 | thunar-volman is the volume management bit for thunar. | 09:37 |
amesser | have installed thunar and thunar-volman but usb mounting does not work | 09:40 |
amesser | "unsupported media usb-storage" | 09:40 |
amesser | thunar can also use gvfs :-) | 09:40 |
Unit193 | Yes it does make use of gvfs. | 09:43 |
amesser | Is there any non-KDE file manager without gvfs which can do usb mounting through udisks | 09:45 |
amesser | I tried some but they all do not show the stick | 09:45 |
amesser | I'm sorry, have to do something else now, only following losely the discussion | 09:46 |
Unit193 | spacefm, perhaps? | 09:46 |
gnarface | maybe nautilus? | 09:46 |
amesser | nautilus is gnome and pulls in gvfs :-) | 09:47 |
Unit193 | IIRC spacefm is a fork of pcmanfm, but I could be wrong, It recommends udisks as the author wasn't fond of GNOME things. | 09:48 |
amesser | hmm, i'll give it a try | 09:49 |
amesser | one last thing, then my employer calls me to work :-) | 09:49 |
MinceR | udisks is a gnome thing | 09:49 |
Unit193 | Ah, my bad. Though, it does recommend udisks2 so I presume that's what it uses to mount. I know it's a little different nevertheless. | 09:51 |
amesser | terra,Unit193: spacefm works without gvfs | 09:53 |
amesser | I installed it an i was able to mount usb stick | 09:53 |
Unit193 | Glad it solved your problem, amesser. | 09:53 |
amesser | well, terra's problem :-) | 09:53 |
amesser | i actually logs to the console thath it calls udisksctl | 09:53 |
amesser | it | 09:53 |
man_in_shack | any virtualbox users in here? | 11:21 |
blinkdog | not so much at home, but I use virtualbox through vagrant at work a lot | 11:22 |
man_in_shack | wat | 11:22 |
man_in_shack | that sounds messy | 11:22 |
blinkdog | haha well, it does take a little getting used to | 11:23 |
man_in_shack | blinkdog: what host/guest you using in that monstrosity? | 11:33 |
blinkdog | Host is a laptop running Mint 18; Guest is typically CentOS 7 because most of our work infrastructure migrated to that | 11:39 |
blinkdog | I'm usually testing something intended to run on a VMware VM running CentOS 7, so I try to test on something close to what it'll be in production | 11:40 |
man_in_shack | fun | 11:40 |
blinkdog | The admins aren't quite brave enough to go Devuan; but I'm working on them | 11:41 |
man_in_shack | stab them until they agree | 11:41 |
terra | anyone using galculator without gtk3? | 13:03 |
terra | simple calculator app pushes 73mb dependency due to gtk3 | 13:03 |
man_in_shack | i use qalc | 13:06 |
terra | man_in_shack: it is 8mb with deps. not bad but I think I could get it through for 1-1.5mb | 13:08 |
man_in_shack | well you could always use bc | 13:09 |
terra | :] | 13:10 |
terra | I installed deb file from Debian wheezy. Total install size is 1.7mb. Not bad. | 13:17 |
man_in_shack | i know them feels | 13:18 |
man_in_shack | i have a nas-type system booting off usb keys with no hdds in it yet | 13:18 |
man_in_shack | 5 dvd drives though | 13:18 |
man_in_shack | (2 of which are bluray, and only 1 of those 2 i actually paid for :P) | 13:19 |
man_in_shack | i have vlc on it in case i want to play a dvd live, without ripping it | 13:19 |
man_in_shack | and managed to get everything else i need without installing gtk :D | 13:19 |
man_in_shack | thanks to dvdbackup | 13:20 |
terra | /usr/share/locales + /usr/share/man = 170mb | 13:20 |
man_in_shack | hehe | 13:20 |
terra | This means I can Install Java | 13:21 |
terra | The very good thing about distros like Alpine, they install packages in very minimum manner. | 13:21 |
terra | documentation and manual pages are separated into different package as default | 13:22 |
man_in_shack | ew java | 13:22 |
man_in_shack | holy fuck vbox is fast now | 13:22 |
terra | This is going to be an office client. They need java. | 13:23 |
man_in_shack | turns out i screwed up stuff with my kernel by not going through major settings, or possibly just the upgrade from 4.18.11 to 4.18.15 | 13:23 |
man_in_shack | also host i/o cache made a huge difference | 13:24 |
man_in_shack | but 3d acceleration REALLY sped up this shitty win10 guest | 13:24 |
rsevero | Hi. I'm trying to install ndjbdns (http://pjp.dgplug.org/ndjbdns/) in a devuan machine and possibly make a deb package of it. "./configure" is complaining with the following error: | 15:31 |
rsevero | ./configure: line 3751: syntax error near unexpected token `[systemd],' | 15:31 |
rsevero | ./configure: line 3751: `PKG_CHECK_EXISTS([systemd], [have_systemd=yes], [have_systemd=no])' | 15:31 |
rsevero | I don't know if this issue should be solved on the upstream side or on Devuan side. And what should be done. Ideas? | 15:32 |
fsmithred | rsevero, is there a configure option for --with or --without-systemd? | 15:32 |
rsevero | Not that I can find it. I believe the developer was expecting that the availability of systemd be treated automatically. | 15:34 |
fsmithred | that could be. If the option exists, it should be listed in ./configure --help (but no guarantees) | 15:36 |
rsevero | Yeah. That's were I just looked for it. Nothing... | 15:36 |
fsmithred | if the error message is accurate, there really is a syntax error in the code. | 15:36 |
fsmithred | check it and fix it if you can | 15:37 |
rsevero | I searched the web a bit to understand why PKG_CHECK_EXITS call if resulting in a error. Couldn't find why. | 15:37 |
rsevero | I tried that. I will search more and try again. | 15:38 |
fsmithred | I found this: | 15:39 |
fsmithred | PKG_CHECK_EXISTS(MODULES, [ACTION-IF-FOUND], [ACTION-IF-NOT-FOUND]): Checks to see whether a particular set of modules exists | 15:39 |
fsmithred | so look at the code and see what ACTION-IF-NOT-FOUND does, and maybe change the action | 15:40 |
debdog | is systemd involved in the kernel module loading process? like, do files in /etc/modprobe.d/ work for blacklisting? | 15:47 |
debdog | hmm, ok, prolly the wrong place to ask | 15:48 |
djph | debdog: it's not part of the kernel ... yet | 15:53 |
debdog | ok, thanks | 15:54 |
Digit | does the new firefoxen work ok for others? often i find myself waiting between 4 to >40 seconds between just basic interface clicks before it responds. | 18:17 |
Digit | idk how much it's because i no longer have policeman addon and others trussing up the insane bloat on pages. | 18:21 |
buZz | trussing? | 18:22 |
Digit | to tie, bind, or fasten. yeah. would that have read better if i said, perhaps, blocking? | 18:24 |
buZz | Digit: blocking sounds more like you dont want to support them ;) | 18:29 |
Digit | policeman let me manage various things, blocking or allowing. so it wasnt all just blocking. hence my opting for another word. trussing seemed to do it. but this is all burrying the lede, missing the pertinent matter at hand... | 18:30 |
Digit | newer firefox... is it slow for everyone or just me? | 18:30 |
* buZz tries http://webglsamples.org/aquarium/aquarium.html in latest stable firefox | 18:31 | |
buZz | works fine | 18:31 |
buZz | 20fps with 10000 fish | 18:31 |
buZz | anything specific you see slow speeds on? | 18:33 |
Digit | well, whatever it is that's slowing, it's nothing whatever this is doing. | 18:34 |
Digit | i wonder if it's the ublockorigin that's contributing to the slowness. | 18:34 |
Digit | anything specific? like i said, often i find myself waiting between 4 to >40 seconds between just basic interface clicks before it responds. any and all. | 18:34 |
buZz | weird, nothing like that here | 18:35 |
buZz | you installed proper gpu drivers? afaik firefox leans quite heavy on GPU nowadays | 18:35 |
Digit | so... it is just bloat? :( | 18:35 |
buZz | ? no its using the GPU to render faster | 18:36 |
buZz | i think you can disable it in options somewhere | 18:36 |
buZz | why not answer the question though? | 18:36 |
Digit | and it needs this because.... bloat. didnt have a problem with firefox back 14 years ago. XD | 18:36 |
buZz | 14 years ago???? | 18:36 |
Digit | what question? | 18:36 |
buZz | the one with the ? behind it | 18:37 |
Digit | oh, drivers. yeah, just recently. | 18:37 |
Digit | it looked rhetorical with what you said after, sorry. | 18:37 |
buZz | so maybe you installed them incorrectly if you have low performance now | 18:37 |
furrywolf | firefox is substantially larger than a complete fully functional linux server install. | 18:37 |
furrywolf | so... bloat. lol | 18:37 |
Digit | if it were graphics, methinks that fish thing woulda struggled more. | 18:37 |
buZz | furrywolf: how are a webbrowser and a OS installer comparable? | 18:38 |
buZz | Digit: its not suffering for me, is it for you? | 18:38 |
Digit | it? suffering? ? | 18:38 |
buZz | yes | 18:38 |
buZz | 'that fish thing' that you just referenced | 18:38 |
Digit | that doesnt help clarify your question | 18:38 |
buZz | you never responded with your performance on that link | 18:39 |
furrywolf | I didn't say "installer". | 18:39 |
Digit | buZz: you seem to be asking questions about what i've already stated, and it's getting exhausting. | 18:39 |
furrywolf | firefox is larger than an entire installed OS with standard utilities. | 18:39 |
buZz | Digit: and you seem to just want to complain and not get to the core | 18:39 |
buZz | furrywolf: thats just wrong | 18:39 |
buZz | furrywolf: firefox is ~60MB , a full OS a couple GB | 18:39 |
furrywolf | I have a Debian install in about 60MB. :P | 18:40 |
buZz | furrywolf: cool, go use it for rendering webgl pages then :P | 18:40 |
furrywolf | arm, runs my weather station. | 18:40 |
nemo | buZz: well technically (I just checked w/ dpkg -x) firefox burns 157MiB these days unpacked ☺ | 18:44 |
buZz | nemo: :D | 18:45 |
nemo | 41MiB dpkg | 18:45 |
buZz | yeah, i ment the .deb :P | 18:45 |
nemo | buZz: but... yeah. what a browser is expected to do keeps climbing | 18:45 |
buZz | either way, i see really no issues with its performance | 18:45 |
nemo | everyone keeps throwing crap in there - thankfully chrome's webusb is currently in disgrace | 18:45 |
buZz | hehe, i was laughing hard at webmidi :P | 18:45 |
nemo | looks like the core libxul.so which does most of the actual browser is 93MiB these days. hm | 18:46 |
buZz | regardless, its really recommended to upgrade browsers more often then once per 14 years | 18:46 |
furrywolf | meanwhile, they just removed RSS support, because it's an evil thing you shouldn't use. | 18:46 |
nemo | furrywolf: well. more like people weren't using it. annoying as all hell. ofc google is kinda to blame there | 18:47 |
nemo | took over RSS market, then killed off their reader | 18:47 |
nemo | RSS is not very popular amongst the web "publishers" - you can't put ads in it | 18:47 |
buZz | i use RSS all the time, to find new torrents ;) | 18:47 |
nemo | so not surprising they worked hard to get 99.99% of users to stop using it | 18:47 |
nemo | furrywolf: what frustrates me even more than removing live bookmarks (basically useless) and the default RSS feed renderer (WTF guys, that thing was only a simple XSLT transform file!) | 18:48 |
nemo | furrywolf: was removing the hooks | 18:48 |
nemo | which makes it harder for addons | 18:48 |
nemo | managers... | 18:48 |
nemo | hm. let's see what happens if I click on an RSS feed in nightly | 18:48 |
furrywolf | I'm a supporter of the "someone paid us to remove it" theory. | 18:49 |
nemo | prompts for external reader. boo | 18:49 |
Digit | links looks more and more appealing, every firefox upgrade. this latest one, especially. | 18:50 |
buZz | fyi, NCSA Mosaic works well on modern linux | 18:50 |
buZz | https://github.com/alandipert/ncsa-mosaic | 18:51 |
buZz | havent seen a faster x11 webbrowser | 18:51 |
nemo | Digit: my fav tmux-friendly browser is w3m ☺ | 18:51 |
buZz | w3m even can do images in terminal :) | 18:52 |
nemo | I currently have 19 tabs open in w3m | 18:52 |
nemo | buZz: yeah. long time ago I did in fact do my entire session in a vt ☺ | 18:52 |
buZz | nemo: are you using w3m-img ? :) | 18:52 |
buZz | i miss my old vt520's :( | 18:53 |
nemo | I have a screenshot from then of w3m w/ the image rendering on my customised console | 18:53 |
Digit | i remember when firefox wouldnt show signs of slowing down until over 1400 tabs. then one upgrade that became 900. then another, 500. skip to today... seems anything in double figures, or not even. | 18:53 |
nemo | but... I dunno. I don't really find the function that useful most of the time, and esp in my tmux session, so I don't really use it anymore | 18:53 |
nemo | when I want to view an image I just load it explicitly in cacaview | 18:53 |
nemo | Insert→i | 18:54 |
buZz | i just use the ungoogled chromium with uBlock and tab memory saver | 18:54 |
buZz | got like 10 windows with >100 tabs each , not even using 4GB total | 18:54 |
* Digit rolls his eyes so hard the table flips | 18:55 | |
buZz | Digit: you found the reason for the 40second click delay yet? | 18:55 |
Digit | nope. just found someone who pretending to help was giving me bum info. | 18:55 |
buZz | oh ok | 18:56 |
buZz | was he by any chance operator in #devuan? :P | 18:56 |
* buZz coughs | 18:56 | |
buZz | Digit: which firefox version introduced the 40second click delay on your computer? | 18:57 |
Digit | 60.2.2esr | 18:58 |
Digit | was a pre-quantum version prior. | 18:58 |
buZz | so, last weeks version? if you go back to previous it removes the delay? | 18:58 |
Digit | but like i mentioned, i wonder if policeman was doing most of the bloat-removal. | 18:59 |
buZz | i have no plugins in firefox and no clickdelays, so doesnt sound like it | 18:59 |
Digit | how do i go back to previous. i tried n failed to find how to do that in devuan. | 18:59 |
Digit | s/previous./previous?/ | 18:59 |
Digit | buZz: how many tabs do you have open in firefox? just the one to check? sounded like you dont even use it. | 19:00 |
buZz | i had one to check , as i said previous i use chromium usually | 19:00 |
Pause | Thank You to Devuan for existing. I was already wary of systemd, but after an "upgrade" to Ubuntu 18.04 from 16.04 hosed my VPS of things that had worked perfectly fine until then, I decided to try Devuan. Things are instantly more manageable. | 19:00 |
buZz | Pause: welcome :) | 19:01 |
Pause | :) | 19:02 |
Digit | apt-cache showpkg firefox just shows the version i have, nothing older. ... i suppose i need to add some older repos then? :3 | 19:04 |
buZz | the firefox-esr in devuan is just debian's , so you could grab older versions from http://ftp.debian.org/debian/pool/main/f/firefox-esr/ | 19:06 |
Digit | thanks. :) | 19:06 |
nemo | Digit: are you having UI performance issues? | 19:33 |
Digit | we shall see, once i close this firefox and restart it back in version 52.9. | 19:33 |
nemo | Digit: there's a couple of recent firefox changes that screw over some older setups - I encountered one of them with my coworker's xrdp and with my ssh -YC which was that they'd disabled the X acceleration - that can be turned back on in settings... | 19:34 |
nemo | Digit: the other thing to look into is whether gl acceleration is working at all and, if it is, hopefully not in software emulation | 19:34 |
nemo | I can find that first setting if you're interested though | 19:34 |
Digit | hrm. 52.9.0 seems to have the same chunky ui. maybe i didnt go back far enough? ~ maybe just need to re-enabled some addons, n disable the newer ones added. ~ ah, nope, cant re-enable. | 19:38 |
buZz | yeah the GL was my first hunch nemo | 19:38 |
* Digit goes back another version to see if that takes it back to where it's pleasant for him, before revisiting whether to bother with the pain of the new version | 19:39 | |
nemo | Digit: hang on lemme find that value. I have it in one of my profiles for ssh -YC ☺ | 19:49 |
nemo | Digit: https://bugzilla.mozilla.org/show_bug.cgi?id=1263222 | 19:54 |
nemo | gfx.xrender.enabled;true | 19:54 |
* Digit sets to true | 19:55 | |
nemo | Digit: also check your gl acceleration values in about:support and whether your gl acceleration is functioning in glxinfo | 19:59 |
nemo | Digit: I learned on here, eventually, that AMD cards need to install a non-free package to function | 19:59 |
nemo | and this is not super obvious anywhere but dmesg | 19:59 |
Digit | got proprietary nvidia reinstalled recently. | 20:00 |
Digit | checking on those values you mention | 20:00 |
nemo | Digit: BTW, if those are not working you might possibly want to disable gfx.xrender.enabled once you get them working, unless you use firefox in remote X | 20:03 |
nemo | Digit: oh. also be sure to try a clean profile just in case you have something crazy set in your gl config settings from long ago (I've done this) | 20:04 |
nemo | for example layers.acceleration.disabled;true or gfx.content.azure.backends set to something odd | 20:05 |
buZz | Digit: what framerate did that fish thing give you with 10000 fish? | 20:05 |
buZz | also 52.x was released 4 months ago, did you have this delay 4 months ago aswell? | 20:07 |
nemo | er wait what? 52 is only 4m old?? | 20:08 |
buZz | the debian package .. | 20:08 |
buZz | ~ 2018-06-27 | 20:08 |
nemo | ah | 20:08 |
nemo | firefox 52 on their calendar was released in 2016 | 20:08 |
buZz | so eh, 52.9.0esr release | 20:09 |
buZz | i'm still under impression its not firefox thats causing this slowdown at all | 20:09 |
nemo | hm no 2017-03-07 Firefox 52 - I misread | 20:09 |
buZz | which GPU is in this system btw? | 20:10 |
KatolaZ | wow, one year old | 20:10 |
KatolaZ | must be so damn wrong.... | 20:10 |
KatolaZ | (kidding ;P) | 20:10 |
Digit | nemo, in about:support, under graphics, features: compostiting, basic; decision log hw_compositing "blocked by default: acceleration blocked by platform" & opengl_compositing "unabailable by default: hardware compositing is disabled". glxinfo says a lot of glx stuff and "direct rendering: Yes". | 20:10 |
Digit | buZz, prior attempt, 30. after setting gfx.xrender.enable to true and running this older version, 14-16. | 20:11 |
buZz | blocked? :O | 20:11 |
buZz | Digit: does glxinfo say its using NVIDIA's driver, or mesa's ? | 20:11 |
Digit | glxinfo | grep mesa returns nothing. | 20:11 |
buZz | ActiveYes | 20:12 |
buZz | DescriptionNVIDIA Corporation -- GeForce GTX 1060 3GB/PCIe/SSE2 | 20:12 |
buZz | in my firefox's about:config | 20:12 |
buZz | and also; | 20:12 |
buZz | HW_COMPOSITING | 20:12 |
buZz | blocked by default: Acceleration blocked by platform | 20:12 |
buZz | OPENGL_COMPOSITING | 20:12 |
buZz | unavailable by default: Hardware compositing is disabled | 20:13 |
buZz | Digit: whats the > server glx vendor string: | 20:14 |
buZz | in glxinfo ? | 20:14 |
buZz | should be fourth line from the top | 20:14 |
nemo | also maybe... glxinfo | grep renderer string | 20:14 |
nemo | er | 20:14 |
buZz | buzz@h81m:~$ glxinfo | grep renderer | 20:15 |
buZz | OpenGL renderer string: GeForce GTX 1060 3GB/PCIe/SSE2 | 20:15 |
nemo | glxinfo | grep "renderer string" | 20:15 |
nemo | yeah taht | 20:15 |
nemo | ☺ | 20:15 |
buZz | :D | 20:15 |
nemo | Digit: both mozilla and google maintain blacklists of gl under linux based upon strings - these lists are very unreliable and both vendors block things they didn't intend, the lists aren't even synchronised despite often blocking for same reasons | 20:16 |
nemo | Digit: if you are sure that your gl accel is enabled (check that renderer thing above) then you can force-enable it in firefox config | 20:16 |
nemo | worst that happens is maybe it screws up or crashes and you just have to disable it again | 20:16 |
Digit | buZz: under gfx.crash-guard.glcontext.* i see 4.5.0 NVIDIA 375.39 and a Geforce GTX 750/PCIe/SSE2, if that's the right place. glxinfo's vendor, NVIDIA Corporation. renderer GeForce GTX 750/PCIe/SSE2. | 20:16 |
buZz | sounds right, yeah | 20:17 |
nemo | layers.acceleration.force-enabled;true | 20:17 |
buZz | gtx750 isnt that slow, should be fine | 20:17 |
nemo | webgl.force-enabled;true | 20:17 |
buZz | i still dont understand how this could introduce a 40 second delay on clicking on webpages | 20:18 |
buZz | isnt the system just out of ram, or something? | 20:18 |
nemo | Digit: what are you using for your window manager? does it use compositing? | 20:18 |
nemo | Digit: my boss had an old mac - he upgrades to latest OSX and all of a sudden lag eeeeeverywhere | 20:18 |
nemo | it's 'cause they force-enabled compositing on latest OSX and his system only had a modest amount of video ram | 20:19 |
nemo | your card has a gig which is respectable, but a large display with a lot of stuff being composited can suck up a lot too | 20:19 |
nemo | then you add on the browser which wants a lot for page acceleration... | 20:19 |
Digit | 3gig left of my 32. xmonad, no compositing. | 20:19 |
nemo | Digit: um. your card has a gig, not 32 gigs, at least in standard config ☺ | 20:20 |
nemo | Digit: vRAM that is | 20:20 |
Digit | just firefox that becomes unresponsive, terminals, emacs, whatever other guis, mpv, all fine. | 20:20 |
nemo | Digit: did restarting firefox with gfx.xrender.enabled set to true help? | 20:20 |
buZz | Digit: and other QT apps? | 20:20 |
nemo | Digit: and did you try a clean profile? | 20:20 |
buZz | firefox is a QT app after all | 20:20 |
nemo | buZz: no it isn't? | 20:21 |
nemo | gtk | 20:21 |
buZz | its not? :O | 20:21 |
nemo | by default anyway | 20:21 |
nemo | there are qt builds but it is non-standard | 20:21 |
buZz | ah, they stopped doing that in 2010 | 20:21 |
Digit | nemo: i know that's not my gfx card's memory. i was answering "<buZz> isnt the system just out of ram, or something?" | 20:21 |
nemo | ah | 20:21 |
nemo | ok | 20:21 |
Digit | oh, need to restart for that to take effect... 1 moment. didnt realise. | 20:21 |
buZz | less that 15% ram available does come close to 'full' | 20:21 |
buZz | than* | 20:22 |
nemo | Digit: you should always restart after changing any of the options mentioned above. and definitely test in a clean profile | 20:22 |
nemo | Digit: firefox -no-remote -P | 20:22 |
nemo | Digit: both avoids screwing up your main profile and eliminates any possible weird settings you might have - or extensions | 20:22 |
nemo | but yeah. most of the graphics optiosn require restart | 20:22 |
nemo | since that's all setup on launch | 20:22 |
Digit | should that be --no-remote? | 20:23 |
nemo | no | 20:23 |
nemo | or at least it doesn't matter ☺ | 20:24 |
nemo | https://developer.mozilla.org/en-US/docs/Mozilla/Command_Line_Options | 20:24 |
nemo | looks like they accept both these days though | 20:24 |
Digit | bbialb, food/toilet/break. | 20:25 |
Digit | well, so far so good (added stylish, classic theme restorer & policeman, and set gfx.xrender.enabled:true)... lets see how it gets once i add half a dozen pinned tabs and whatever other essential addons. who knows what cruft was weighing down my old profiles. a lot, i suspect. | 21:05 |
* Digit makes this action just to remind himself to apt-pin firefox-esr once he's done faffing around with the addons, so he doesnt suffer pains next apt-get upgrade. | 21:06 | |
furrywolf | yeah, and apt-get upgrade shouldn't do things like uninstall all your extensions and all the other shit firefox has been doing lately. a normal upgrade should only make bug fixes... | 21:08 |
furrywolf | firefox seems to be permanently unstable now. | 21:08 |
nemo | the problem for debian is that their LTS does not match Firefox' | 21:14 |
nemo | and they don't have the manpower to backport patches themselves | 21:15 |
nemo | and, yeah, firefox chose to remove their entire extension architecture | 21:15 |
nemo | furrywolf: the main one that hurts for me is tab panorama | 21:15 |
nemo | the others I've adapted too | 21:15 |
nemo | and someone just closed the main tab panorama bug as "fixed" despite lots of functionality still missing - all the existing implementations are basically unusable | 21:15 |
furrywolf | firefox is rapidly becoming shittier. | 21:22 |
nemo | I'm definitely not thrilled about the webextension move either | 21:24 |
nemo | nor the forced signed extensions | 21:24 |
nemo | although thankfully that's not in beta and nightly | 21:24 |
nemo | was supposedly a security thing - people blaming firefox 'cause random malware was cramming in browse extensions or the extensions were badly written and were crashing browser. | 21:25 |
nemo | but man is it irritating to lose all my awesome customisations | 21:25 |
nemo | especially tab panorama which I really am addicted to | 21:25 |
nemo | furrywolf: I'm still running firefox 52 in one profile 'cause I don't want to give that up | 21:25 |
nemo | other profile is on nightly so I can do whatever the hell I want to w/ the existing limited extension system | 21:26 |
furrywolf | mozilla also has developed a serious attitude problem lately... every time there's any dispute over anything, their response is basically "we're in charge, the community doesn't matter, go fuck yourselves if you don't like it." | 21:31 |
furrywolf | bbl, time for work | 21:39 |
* Digit getting things back to sanity. classic theme restorer, policeman, global dark style (with nightman.css added to global in its preferences), adblockplus, ~ trying to remember what else is important | 21:40 | |
Digit | thanks for the help. and sorry for snippyness/miscommunications. i was not handling the stress well. | 21:41 |
Digit | furrywolf: yep. well depicted. sorta become a defacto monopoly and the maladies that come with it. | 21:42 |
Digit | buZz: that appology for snippyness/miscommunications's particularly for you. sorry 'bout that. :) | 21:43 |
buZz | <3 dont get stressed | 21:43 |
buZz | :) | 21:43 |
* buZz throws weed at Digit | 21:43 | |
* Digit clicks fingers and points, mutters "i knew i was missing something else!" and reaches for the bong | 21:44 | |
buZz | ;) | 21:45 |
buZz | did you fix the issue eventually? or just accepted as new norm? | 21:45 |
ServiceRobot | good afternoon gentlemen | 21:46 |
Digit | current firefox setup, 52.9.0, just 4(!) addons, and yet to test with large number of tabs. things are looking promissing though... better than the new fatfox. | 21:48 |
buZz | Digit: so you didnt find what caused it? too bad | 22:00 |
Digit | chalking it up to comorbid mystery glut of cruft in long-abused profile, worsened by upgrade. | 22:01 |
Digit | but, we'll see how much improved this is once i get back up to my typical use style. | 22:02 |
buZz | ah, with removing the profile it disappeared? | 22:03 |
Digit | yeah, fresh profile, with policeman & custom theme restorer, no lingering noticeable delays persisting. but, like i say... this is not yet conclusive until i get back up to many tabs. | 22:04 |
buZz | yeah, would be curious to see the same test on current firefox aswell | 22:05 |
buZz | also with measured click-response times | 22:05 |
Digit | yeah. i am curious in that too... but it's just too painful to warrant going back to. wont be able to use policeman, which is the biggest pain. | 22:06 |
buZz | i guess, only browser plugin i use are uBlock origin and Tab Memory Saver | 22:07 |
Digit | sorry to put my convenience above our scientific curiosity. | 22:07 |
buZz | Digit: i just ask questions to try and help all | 22:07 |
buZz | like, if tomorrow someone comes with same issue | 22:07 |
buZz | i dont like to tell them 'ask Digit' | 22:07 |
buZz | but i can now tell them 'delete your profile' | 22:07 |
Digit | i didnt delete, so much as just moved to a new one. the old's still there if this was of no use. | 22:08 |
buZz | same thing | 22:08 |
buZz | avoid using it :P | 22:08 |
* Digit whimsically contemplates sending bug report to mozilla, stating that the latest firefoxen caused panic attacks, rage outbursts, and other stress maladies. | 22:14 | |
fsmithred | and gunshot wounds to the computer | 22:14 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!