<jlicht>is the intention to just ignore these clearly very angry people on the ML, or what? <jlicht>or alternatively, get some kind of grey-listing system enabled on the ML for the next days perhaps <ng0>gnu_srs2: there are now 6 maintainers (or something around that) <jlicht>gnu_srs2: it is the middle of the night for a lot of guix users, so I do hope we are still at a stage of civil discourse where we can at least give each other the benefit of healthy sleeping habits :-) <verisimilitude>It's a shame I no longer feel at ease with using Guix as my GNU/Linux distribution. <verisimilitude>There's no longer the expectation that someone won't add something malicious for tangential purposes. <verisimilitude>My Guix has been unable to successfully update for months, anyway. <verisimilitude>This would change if the current Guix administration is removed, though; I've tried contributing to this project before, admittedly accomplishing nothing, but I certainly won't even try to help again until something is done with regards to this vile nonsense. <mange>Can you explain the logic by which you're getting to "Guix maintainers might modify Guix to install malicious software on my machine"? <verisimilitude>Since tangential political concerns are being raised, I'm no longer expecting that malicious software of some sort, to stop ``bad behavior'', wouldn't be added, perhaps. <nckx>verisimilitude: That makes no sense. <nckx>That would go against everything GNU stands for, I mean. <truby>verisimilitude: isn't this the whole point of free software? you have the entire source available to see how guix installs things and what it installs... <verisimilitude>This was already one of the more tangentially political GNU projects, but this recent drama has made it seem worse. *truby decides on reflection to not get involved and go back to reading sicp <verisimilitude>What I mean is that I no longer believe the Guix administration could be trusted to follow what GNU stands for. <truby>I hope to become more wise as I read it :-) <mange>What does GNU stand for? (Principally, not the letters. :P ) <mange>What has the Guix project done to make you doubt its commitment to that? <nckx>mange: Ask 10 people & get 11 answers, I suspect, but freeing & empowering computer users will be the core of it. <mange>I do know that, but I don't understand how that statement causes you to doubt Guix's commitment to "software freedom for users of software". Could you help me to understand that progression? <nckx>verisimilitude: That statement is hardly ‘tangential’ to GNU, though. Or incompatible with software freedom. *nckx is also not in the mood for this, goes hackin'. <verisimilitude>By claiming that a core value of GNU is undermined in this way, they actually seek to promote tangential concerns, clearly meaning they seek to advocate for ``minorities'' or other ``oppressed'' people at the expense of others, believing this to be an issue of explicit keeping out rather than lack of interest from those parties. <nckx>There's certainly no logical connection to malware in Guix, which will never be acceptable. <mange>verisimilitude: To check that I understand your objection to that statement correctly: do you object to the spirit of the statement ("The GNU Project we want to build is one that *everyone* can trust to defend their freedom."), or do you object to the claim that "Stallman's behavior over the years has [...] alienate[d] a large part of those we want to reach out to"? <truby>hey, there's a ccls package now \o/ <truby>ty to whoever added that :-) <nckx>truby: That would be Mathieu Othacehe (I don't know their nick, if they even frequent #guix, but at least you can send them good vibes). <truby>I'll have to try it out at work tomorrow :-) I had some hacked together thing I was using internally but this package looks better than mine! <verisimilitude>I believe claiming GNU isn't already for everyone is disingenuous and that RMS has alienated people to largely be a lie. I believe it's a snake seeking to slither deeper inside, because if GNU isn't already for everyone, then that means GNU needs to change to accomodate everyone, and that means taking politically tangential actions to do so. <verisimilitude>Andy Wingo has already made the increasingly common claim that Free Software volunteers are ``privileged''; it's an interesting way to attack people doing charity work, claiming others want to do it, but just can't and need more help. <nckx>I seriously love the alliterative snake imagery. ***catonano_ is now known as catonano
<mange>verisimilitude: I take it you think that GNU is already for everyone. Are there any ways that you think GNU might make it difficult for people to contribute or be involved? And do you think that harms GNU in any way? <verisimilitude>I don't think GNU makes it particularly difficult to contribute, no. There's the issue of copyright assignment that's been cited as bothersome, however. <rk4_>hmm @ wingo's blog, how alarming <rk4_>ah screw it. i don't have the energy for this petty 2019 politics. i'm out. what happened to freedom, tolerance and forgiveness. i don't want to be part of this anti-rms bandwagon. <Gamayun>verisimilitude: Even if RMS were the very model of good behaviour and leadership, I don't think having the same person in a leadership position with no rotation or relief for over 30 is a healthy sign in any organisation... <nckx>r4k_: What's ‘2019’ about it? The exact same problems were already well-known and -discussed in 2009… <Gamayun>As someone who admires RMS, I always thought the talk of "hero worship" was overblown. Seems like it wasn't. <nckx>It's interesting how utterly important it is to them to protect the ‘this is a recent reaction to recent events!!’ spin, when all the evidence contradicts it. Allows pretending that it's all the fault of some random silly Medium post, I guess. Must grasp at that straw, or the carefully constructed rms strawman comes crumbling down. <nckx>Holy mixed metaphors Batman, that was terrible. <nckx>Gamayun: As someone who admires so much of what rms has done through decades of hard and thankless work, I think the fanchildren are his own worst enemies at this point. <Gamayun>Well, he does advocate against US parents coddling and infantilising their children and young. Hopefully he'll be able to let his own "children" grow up too! ;-) <nckx>…and learn that letting them grow up doesn't mean letting go entirely. But it does require trust. <nckx>Gamayun bringin' the actual good metaphors tonight, thank you. <Gamayun>nckx: Sure. ;-) Hope you're not getting too swamped. <nckx>Tonight's been better. Or I just ran out of energy to force myself to take it seriously. Either's good 🙂 <verisimilitude>I don't think it's an issue, mange. I'm not worshipping RMS, merely reacting in a way I find appropriate to this obvious hostile takeover being orchestrated. <verisimilitude>Andy Wingo has previously stated he'd prefer working with someone incompetent, so long as the incompetent shared his political beliefs. Miguel de Icaza is a Microsoft shill who decided it was recently a good time to share RMS stories. The woman who wrote the Medium post is a Salesforce employee who also has ties to defense contractors. <verisimilitude>Corporate interests clearly want to split Free Software into impotent shards and a good first step is introducing tangential political beliefs people can start bickering over, along with attacking others. <nckx>Gamayun: When you're even vaguely representative of a project (or seen as such) there's this weird, self-imposed moral obligation to take critics seriously even while they're calling you a *checks notes* …‘NSA plant’. Once you realise you owe them nothing, things get marginally better. <verisimilitude>Even if RMS has made GNU unwelcoming, Free Software is distributed. <nckx>verisimilitude: Then stop bickering and attacking others? That would actually foil their plan. <verisimilitude>I'm a user of GNU software, not really a contributor, but I work on my own software on my own, how I want to. GNU is not at the heart of this. Anyone who wants to contribute to Free Software but wouldn't want to do so alone at some point is weak, in a way. <verisimilitude>I'm simply raising some concerns about the leadership of Guix, nckx. <nckx>Then they were entirely lost between all the needlessly aggressive rhetoric, which is another very good reason to drop it. Apart from the obvious fact that such invective is not welcomed here, of course. <verisimilitude>Until next time, then; I'll be giving thought to which distribution I want to use next. <Gamayun>And here I thought the original corporate "open source" objection to "free software" was the bickering over "tangential political beliefs". <Gamayun>But I get some of the concern. I hope the alternative to RMS isn't the endless stream of well-presented media-savvy "leaders" that use NGOs and charities as their personal corporate career runway. Not that anyone's proposing that, but good collective leadership is hard to build. *kmicu [impersonating Robin Williams] Gooooooooooood mornin’ Guiiiiiiiiiiiiiiiiiiiiiixxxx! <bdju>I keep running out of disk space trying to update, then I guix gc, and it seems I then have to rebuild what I just built. I'm definitely building racket for the third time now... really quite awful <bdju>I deleted some generations before the gc this time so maybe I have a better shot <bdju>I've got my cpu at 95C and I'm getting occasional mouse lag <bdju>I guess if this fails again (hopefully sooner than 12 hours from now) I'll add racket to the list of stuff to not update <bdju>I keep going to bed with stuff trying to update and waking up to a failure for one reason or another <bdju>I'm thinking if I knew a bit more there would probably have been a less painful path, like if racket really was built, maybe I could've sent the binary to another machine before the gc <bdju>or maybe because none of the updates have finished, I'm not being left with an actual binary... not sure <roptat>bdju: if racket is built but the transaction (upgrade) is not finished, it's not considered live, because it's not reachable from a gc root, so gc will collect it <roptat>What you can do is run guix environment --ad-hoc racket, which will create a temporary root for the gc <roptat>Then guix gc won't collect racket as long as you are in the environment <bdju>interesting, thanks for that tip <bdju>just to clarify, would I thin run the gc command from inside the guix environment or outside? <bdju>not sure if just having the environment exist is enough <roptat>Gc roots are global. As long as the environment exists, there's a temporary gc root <bdju>I've been having locale issues since the recent attempted updates, but I figure maybe those will get sorted out after the updates all go through <bdju>some stuff spits out errors and then works, but rofi won't launch at all currently <Gamayun>Looks like my nextcloud-client package minus qtwebengine plays nice after core-updates was merged! <roptat>Gamayun, mind to send a patch? :) <Gamayun>Will do. Just need to test and see if it needs cleaned up a bit. It's a hasty adaptation of the owncloud-client declaration, so there might be unneeded inputs and such. <ng0>Jean Louis must have so much time. I don't know why some of you keep replying to Jean, this is the same kind of behavior Jean has shown in 2016. <ng0>I bet someone could setup a markovchain bot and keep Jean busy for a long time if they wanted to.. in theory.. <lprndn>hum. I'm currently using a mix of xfce4 and pantheon as my DE under guix system. The thing is that unless I also put mate-desktop-service in my configuration, pantheon's settings don't work (they don't affect the system). FYI pantheon use more or less gnome-settings-daemon to do that. <lprndn>So my question is, does someone has a lead for where to look for a solution. My guess for now, is that's either an env variable missing or a propagated-input as, otherwise, the mate serive shouldn't affect this behaviour <roptat>lprndn, maybe you need one of the dependencies of the mate-desktop-service? <zimoun>ng0: interresting idea :-) A kind of ELIZA. Maybe M-x doctor is enough. ;-) <roptat>lprndn, maybe something link (simple-service 'gnome-settings polkit-service-type (list gnome-settings-daemon)) would work for you <lprndn>roptat: already tried. :/ but thanks for the hints. I'll look for possible packages! <roptat>lprndn, I mean, if mate-service makes it work, then it must add something to your system, and I see that it only adds that polkit service and mate to the system packages <roptat>so if it's not polkit, then it's one of the packages in mate <Gamayun>ng0: Certainly seems like one busy sealion. <lprndn>roptat: hum... maybe dbus or dconf? seems kind of plausible. i'll try. *kmicu is happy for only checking mail once a day. It looks like a fun ahead. <kmicu>(Though good thing about Gnus’ adaptive scoring is that I don’t need to read Jean’s messages since ~2015.) <iv-so>hi! is there any converter from spdx license identifiers to guix ones? <g_bor>I have found two ways to get our qemu fail when load fails. <g_bor>One is platform dependent, but looks easy to implement. <g_bor>The other one is platform independent, but seems to be more convoluted. <g_bor>The platform independent way is what civodul pioneered, trigger a kernel panic, use pvpanic, atttach to the guest qmp and watch for the panic event. When that comes, fail hard. <roptat>iv-so, I think I've seen a spdx->license procedure in an importer <roptat>iv-so, spdx-string->license is defined in (guix import utils) <lprndn>roptat: Thanks for the help. dbus and dconf did the job! Now I need to debug the rest... :D <bdju>oh hey my update finished without error this time <bdju>now I'll apply my manifest and see if I can get the stuff I skipped yet or not <lprndn>How can I add a specific output to my system profile? (for exemple glib:bin) I don't find the correct syntax. <leoprikler>I've sent a patch to guix-patches, but it's not showing up. Can someone help me? <leoprikler>@lprndn: Packages have an `outputs' field, which is a list of strings <leoprikler>If you want a package with outputs "out" and "bin", you just specify the field as (outputs (list "out" "bin")) <bdju>oh wow, and rofi really did start working again after those updates <bdju>sometimes computers aren't so bad... <shrdlu68>If I install packages via `guix install` and then do `guix system reconfigure`, are my manually installed packages retained? <shrdlu68>What about `guix system in config.scm /`? <leoprikler>not quite sure about that, but I'm pretty sure that "init twice" is an error <leoprikler>so I don't think there are any guarantees if you do that <shrdlu68>I use kitty terminal emulator, and it's terminfo is not in the ncurses database. Solving this is a simple as downloading it's terminfo and placing it in $TERMINFO_DIRS. What is the best way to solve this? <shrdlu68>I'm thinking perhaps adding this file to skeletons might be the most straightforward solution, just place it in ~/.terminfo/ <leoprikler>call it "kitty-with-terminfo", place the terminfo in it's /share/terminfo <shrdlu68>leoprikler: thanks, looking at how to write packages now. <leoprikler>@shrdlu68 btw. looking at `guix build kitty`, the package already has a /share/terminfo ***ng0 is now known as Guest59160
***ng0_ is now known as ng0
<shrdlu68>So the only thing it lacks is placing it's /share/terminfo in $TERMINFO_DIRS? <shrdlu68>leoprikler: By the way I'm trying to the get the terminfo on a server where kitty is not installed. On my desktop $TERMINFO is `/usr/lib64/kitty/terminfo/`, so it's probably exported by kitty itself. <shrdlu68>Nor does installing kitty on the server help. <leoprikler>doing `echo $TERMINFO` on guix yields `/gnu/store/*-kitty-*/lib/kitty/terminfo` <leoprikler>okay, so what's the issue you're trying to solve <shrdlu68>When I ssh into a server, kitty's terminfo isn't in $TERMINFO_DIRS. <leoprikler>so you can't start ncurses applications on servers you've ssh'd to <leoprikler>I guess that's a problem with your SSH setup then <leoprikler>because SSH won't look for programs, libraries etc. on your local machine, but on the remote <leoprikler>by the way, I can use htop through ssh on kitty without a problem, even with $TERMINFO unset <shrdlu68>I can't, `reset`, for example, tells me "unknown terminal type, xterm-kitty". <leoprikler>1. install kitty (or just its terminfo) on the server <leoprikler>2. when using a server without kitty support, that you don't have control over, switch to a dumber terminal <g_bor>I just tried to install a new machine. <g_bor>Then I guix pulled, and tried a sudo -E guix system reconfgiure, which died on trying to install nss-certs on a locale problem. <g_bor>Do we know how others solve this? <g_bor>I tried the suggestion printed, but it did not work. <shrdlu68>leoprikler:Thanks, I'll option 1, and placing kitty terminfo in skeleton. <g_bor>I solved it by temporarily commenting nss-certs. <leoprikler>@shrdlu68 instead of modifying stuff in /etc and where not, you can also adjust your user .profile <leoprikler>lprndn: did you fix your problem with package outputs? <olivuser>Due to an error on my part, guix system reconfigure wrote to another disk on which there was data dear to me. I already used ddrescue to clone said harddrive, but now dont really know how to proceed. can anyone tell me where to go from thereon? <g_bor>olivuser: it very much depends on what has been overridden. When did you detect the problem? How much data did you have? How much data was written? <g_bor>Was the original partition label a gpt? <g_bor>Was only one partition involved? <lprndn>leoprikler: no, not yet. I just installed it in my user profile for the meantime. <lprndn>leoprikler: no, it's ok. In command line, `guix install glib:bin` does the job. :) <olivuser>g_bor: let me provide you with as much info as I've got. I had another operating system installed on a ssd drive using the lvm2 filesystem (/dev/sdaX). from guix, I system reconfigure'd to the same slot (/dev/sdaX). Gparted says that the filesystem is still lvm2, even though guix would normally have used another file system architecture, and the reconfigure was not successful. I reconfigured yesterday only to discover the <olivuser>problem today. Well it was not a ton of data (GB-wise), but stuff that I use regularly (like, my emacs config file and work-related files). How much data was written I cannot really tell, since gparted displays the partition, but not how much space is free/used. As far as I know, its origional label was not gpt. and the drive which I "overwrote" consisted of more than one partition, but only one of the two is relevant to me. <olivuser>the other partition is only the bootloader which I dont consider critical. <olivuser>and as I said, I already ran a ddrescue of said partition (/dev/sdb2) to another partition which was completely freed of data (/dev/sda5). There were no problems in the process, but if I understand correctly, I have rescued, not yet recovered said data. <leoprikler>olivuser: can you mount /dev/sda5 and inspect the rescued data? <olivuser>"mount: /media/festplatte: unbekannter Dateisystemtyp „LVM2_member“." <olivuser>lol, sorry, it is in german. It says that "LVM2_member", the apparent filesystem type, is not known to mount <olivuser>leoprikler: it seems like I cant, unfortunately, because the filesystem type is not the correct one <leoprikler>It appears mount does not play nice with logical volumes <olivuser>leoprikler: seems that way, it is not listed along the supported fs types <leoprikler>what's the output of `lvdisplay | grep "LV Name"`? <olivuser>I cant find the command lvdisplay, and guix install lvdisplay says the package is unknown <roptat>I don't think guix will change a partition type by itself, and it seems to have detected that it couldn't do anything with it, so maybe you're fine? <olivuser>roptat: I'm unable to boot either the guix from which I reconfigure'd nor the other operating system (solusos). in both cases I'm thrown to grub rescue <olivuser>leoprikler: I see. I ran without grep and it shows two logical volumes, "Root" and "Swap" <roptat>so grub was installed with a faulty configuration? <olivuser>roptat: unfortunately, I'm not sufficiently into bootloaders to be able to judge. probably, since either distribution has its own bootloader. <olivuser>that is, I choose which harddrive to boot from, and then either fires up its own bootloader <olivuser>leoprikler: and the status says NOT available in both cases <leoprikler>Try `modprobe dm-mod` followed by `vgchange -ay` to make them active <olivuser>leoprikler: "Cannot activate LVs in VG SolusSystem while PVs appear on duplicate devices." thats the error message that it throws. It is probably because I had ddrescue clone /dev/sdb2 to /dev/sda5. I have unmounted /dev/sda5, but the error persists <olivuser>"WARNING: PV QmLGP6-ZGqI-oXZJ-FC01-3anB-41Dy-lG1e3i on /dev/sdb2 was already found on /dev/sda5." maybe this warning clarifies <olivuser>/dev/sdb2 586061 468860174 468274114 223,3G 8e Linux LVM <olivuser>leoprikler: nope, its type is only "linux" <leoprikler>okay, so now I want to find out, whether sdb2 is a logical subvolume of sdb1 <leoprikler>do you have GNOME Disks or anything similar to it? <olivuser>I have it opened, what do you need to know? <olivuser>how would I know it is a subvolume? in the program, it is shown as two separate partitions <bdju>when building ghc I seem to get to 100% but then it stays there... shouldn't it say 99% if it isn't really done? <olivuser>w.r.t doesnt ring a bell, unfortunately. There is no mention of extended partition, they are displayed as "partition 1" and "partition 2". Also, Parition one (/dev/sdb1) is fs-type ext4 and bootable, while partition two (/dev/sdb2) is lvm2 pv <olivuser>leoprikler: would it make sense to delete /dev/sda5 (clone of /dev/sdb2) in order to test whether pvchange works? Because I did not run into any error while cloning it, and I suspect I could clone it again should pychange not be successful <olivuser>sorry, I was referring to what you wrote earlier: "Try `modprobe dm-mod` followed by `vgchange -ay` to make them active" <leoprikler>I think, it would make more sense to unplug /dev/sdb <leoprikler>since we're trying to check, whether we can access /dev/sda5 <olivuser>since I already have a copy in /dev/sda5 <olivuser>alright, will do. be back in a minate. already a million thanks my man, super kind of you to help me! <olivuser>leoprikler: wow, so vgchange did actually work, and I can access the partition that I cloned <leoprikler>and is the data there what you expected to be rescued <efraim>oops, forgot to close icecat before compiling efl <olivuser>leoprikler: will this setting persist when I shutdown/start my computer? <leoprikler>I'm not sure -- all you did was activate the volume <olivuser>alright. at least I know now how to proceed. and I'm not sad for the operating system, only the data. <fps>hmm, i still can't login to i3 using a fresh install of guix in virtualbox. the login manager screen just freezes. going to the console with ctrl-alt-f1 and back to the login manager just resets the login manager <olivuser>Now to my second problem :D I accidentally guix system reconfigure'd to the wrong hdd. The data I believed lost is now safe-ish. But still, I cannot boot into guix. Being the sole connected hdd, I enter in grub-rescue, because "symbol 'grub_file_filters' not found". Ideas anyone? ***apteryx_ is now known as apteryx
<fps>the login manager has nifty features like a calender, settings for high contrast, etc.. but there's one job a login manager should excel at <fps>that's logging the user in <lprndn>fps: Did you try with another login manager? <rekado_>olivuser: has grub been successfully installed to *that* disk? <fps>lprndn: nah, i just used the default from the graphical installer where i just chose i want the i3-window manager <fps>lprndn: i think it;s just the default from %desktop-services <lprndn>fps: it's gdm. Maybe slim would be a better fit for you? <fps>lprndn: i guess so. 14:25 < houston4444> but I just saw the "Workaround Qt bug" which is probably the source of my crashes ;) <fps>oops, sorry, i don't have my input devices under control here *nckx also recommends slim. <efraim>python2-efl so far has peaked at 5.85 GB of RAM/swap during build <nckx>g_bor: dutchie had the same nss-certs locale problem 2 days ago, but I don't remember how/if they solved it. This bug refuses to die ☹ <g_bor>nckx: I simply commented it, reconfigured, added it back in and reconfigured again. <nckx>g_bor: and uncommented? Or is it stil borken? <g_bor>no, it is not broken any more. <g_bor>The problem is that the user guix and the system guix gets out of sync on the glibc version. <g_bor>So once you could reconfigure the system and reboot to make the daemon see the new locales it works. <g_bor>you can safely add the the nss-certs back. <g_bor>It is more of an inconvinience. <g_bor>It might be problematic if other things on your system depend on the certs, though <g_bor>most of my systems are near bare-bones. <lprndn>by the way, I sent some patches to guix-patches adding a lightdm service some time ago. I think the multiple patches are a little daunting to review. Is it ok (or even useful) if I resend them separetely? <g_bor>dannym also reported this issue. <nckx>g_bor: Aha, wow, that's bad. <roptat>g_bor, as the news entry says, you should add the locales for the previous glibc in your config <roptat>well, now it's not useful anymore <nckx>I count 3 people in fewer days that somehow missed ‘guix pull --news’ entirely, none of them noobs by any stretch. <roptat>I think it should be shown by default <nckx>g_bor: Did you get a ‘hey you, read our new cool news!’ message at all? (I mean, obviously you would have done it then, but I have to ask. Or maybe you can scroll back.) <apteryx>yeah, perhaps an abbrieved news output could be shown after 'guix pull', with the suggestion to run 'guix pull --news' for more. <nckx>apteryx: I thought we already showed ‘Run guix pull --news’ (in green, IIRC). Problem is, I guix pulled *in order* to test the cool news feature so I didn't pay much attention 🙂 <dutchie>nckx/g_bor: it went away after I tried with --fallback, I think <g_bor>roptat: Could we somehow display the correct hint in the warning about locale? <g_bor>That would be a great help... <nckx>dutchie: Oh, now I remember… I left a message with sneek that they never delivered 😠 Saying that ‘--fallback’ was probably not the fix, just running it until it happens to work. <g_bor>nckx: I will try this again and tell you how it went. <efraim>lprndn: I remember the lightdm patches, I have them starred for looking at later. I'd suggest bumping them with a ping message or something. The only way I'd go for trying for a faster review would be to include a vm config using it <nckx>dutchie: Thank you for filing that bug. I've merged it with dannym's (later) one. <dutchie>ah well, it never showed up on my other computer anyway <g_bor>it seems that only the substitute is mishandled somehow, so that's why fallback helps here. <g_bor>hmm... somehow my other machine insists on trying to build icecat... <nckx>g_bor: I thought so too at first, but then I wonder why it worked for me without --fallback. I used substitutes, and never bothered adding any locales anywhere. 🤔 Oh well. It doesn't really matter which work-around works, folks can just try all of 'em! Yey. <nckx>g_bor: Negative narinfo caching? <g_bor>after a fresh install from the installation media simply guix pull does not show news. <nckx>g_bor: Nor a tip? Interesting (assuming the installer wasn't brand-new; the news is diff-based.) <g_bor>the only tip is to run hash guix <g_bor>this is the installer image from the website <g_bor>I belive the easiest workaroud would be to release a new installer image. Wdyt? <g_bor>Then new users would not immediately bump into this... <nckx>g_bor: Not opposed, but it would only help new users (a strict minority), and current master really isn't in a state that can be released. <g_bor>Do we actually know what is wrong with nss-certs? <nckx>g_bor: I don't know, but nss-certs has been notorious in the past for pushing the boundaries of your average Unix file encoding. Nothing wrong with it, but a very good fuzz test. It doesn't surprise me that it's this package that dies like a canary, again. <g_bor>I believe it is filenames with special characters, but why is it causing a problem? <nckx>Yeah, well, locales, black magic, I stay far away. <lprndn>efraim: Thanks. I'll try do add a vm config soon. But don't worry, it's really not a hurry, I just wanted to know if I could ease the reviewers' job. <nckx>Someone here this week had a bug that went away when they used en_GB instead of en_US. I'm not meddling with that kind of evil. <nckx>Oops, s/file encoding/file name encoding/ above. <nckx>I'm going to pull a guix from Sept 24 and see what happens. Surely it must suggest --news. <leoprikler>Does guix pull even suggest --news if it's the first pull that is done? <leoprikler>by the way, how do I send patches? I've tried guix-patches@gnu.org (some hours ago), but my patch won't show up <nckx>Another good question my poor laptop is chugging away at a VM image to answer 🙂 <nckx>leoprikler: That's it. There's an insanely variable delay. Sorry. <nckx>leoprikler: It's also (human-)moderated on first post from any address. <g_bor>nckx: I just installed both glibc-locales in my user profile, but none helps. <nckx>debbugs.gnu.org is run by GNU, I don't know how many Guixers even have access to its back end. I don't, anyway. <nckx>g_bor: And sources the thing with the stuff? <g_bor>it suggest setting guix locpath, which I have done <nckx>leoprikler: Er. Hm. That's a very good question 😃 <nckx>g_bor: Wouldn't that only affect the user side though, not the daemon (which is doing the unpacking and encountering the error)? <nckx>But you said you already added it to your system configuration, so I don't know. <nckx>- what could be the problem at this point. <g_bor>But how could you add it to the daemon without a reconfigure? <g_bor>I just checked, and building locally works. <nckx>Yeah, I get the catch-22. <nckx>g_bor: herd stop guix-daemon; sudo $SOME/guix-daemon --the-stuff, maybe with LOCPATH magic added. <g_bor>A bit of a sledgehammer solution would be to simply make nss-certs not substitutable, I believe it can be built within reasonable time even on low end devices. <bdju>ghc still stuck on 100% and maxing my cpu almost two hours later... is this normal? <nckx>g_bor: It takes like 3 seconds on my x230t. <g_bor>Anyways, it would be very interesting to see why the substitution fails... This should be some port encoding issue... <g_bor>I have to go now, but will check the backlog if anything comes up. <nckx>bdju: I'm afraid it's not implausible. Could be suspicious if it's been the same ghc process for all 2 hours, but hard to say. What's it building? <bdju>ghc itself as a guix update I mean <bdju>I could report it, but since it doesn't fail I don't really have a build log <nckx>bdju: Build logs are save on success as well as failure, just not printed at the end. <bdju>what about in the case of neither? <bdju>it is stuck building forever <bdju>doesn't seem like a success <bdju>is there a file being written to right now that I can look at? <nckx>Are you sure it's stuck? Why? Could it just take a long time to build? <bdju>it has a progess bar which was at 100% two hours ago <bdju>it looks exactly the same now and my CPU is pegged still ***jonsger1 is now known as jonsger
<bdju>also I don't recall it being a super slow thing to build before <bdju>this is as bad as rust or webkit <nckx>I've never paid attention to GHC build times before. <bdju>I think showing 100% progress for two hours is worse than the stuff that shows the spinning thing without any ETA *nckx is building it now, just to get an idea of what its progress would look like. <nckx>bdju: All nice things have ugly failure modes, that's the way of the world I'm afraid. 🙂 The progress bars do a pretty good job considering their input. <bdju>I spent ages building the gcc toolchain earlier in the same transaction, so I hate to risk losing that progress if I cancel because of ghc <nckx>bdju: If GCC is finished (not building in parallel) it won't be built again. <nckx>Ugh, now the (fast) machine I'm building GHC on is going nuts with ‘@ build-log 22077 160 @ download-progress /gnu/store/ijzn’ leakage. <bdju>oh yeah, I should find my log I guess <bdju>not really sure how to interpret it, but it mush be huge. there's a delay going to the end of the file pressing G in less <nckx>I always use C-c (to cancel line number calculation), > <efraim>I found a breaking bug on the core-update merge! I reconfigured and rebooted, and the battery in my mouse died! <nckx>efraim: This is spooky, because that literally happened to me y'day. <bdju>I feel bad about this but I think I am gonna cancel the build and uninstall ghc. it will be someone else's problem to report. I don't even program in haskell. I had it to run a friend's stuff and maybe learn, but it's just a problem for me at this point. <nckx>bdju: I understand. I just started building ghc@8.6.5 when you said that. I'll let know what happens. <nckx>It's already at [172 of 270]. (guix build has no progress bar). <nckx>Hm. It printed [270 of 270] a while ago but keeps on building a lot of stuff. If this is still the build phase, that would have been the cause of your progress bar 100% right there. <nckx>It doesn't look like a test suite. <sirgazil>To the people who added Blender 2.80, thank you very much :) <sirgazil>I also hear there are more maintainers, and that's great. *kmicu recommends ‘send‑patch more people’ more than ‘bug more people’. <efraim>anyone else seeing gnu/services/cups.scm:208:0: warning: shadows previous definition of `%-location-procedure' at gnu/services/cups.scm:200:0 <sirgazil>kmicu: sure, but patching is not trivial for me. Most things in Guix are beyond my skills :) <nckx>efraim: Not familiar. ‘guix system reconfigure’? <efraim>i'm running make on my guix tree, seeing it for many of the gnu/service files <efraim>i'm pretty sure I ran make clean-go before I started *nckx is making clean-go all now. <kmicu>sirgazil: fair enough, minimal, detailed and reproducible issues are good too ;P <nckx>bdju: There's no mystery (or Guix bug) to GHC, and it wasn't stuck. It prints a misleading progress up to 100% during the first few minutes (that Guix believes) but then continues to build for hours. It's still going. Unfortunate, but not worth adding a hack in Guix IMO. <nckx>Hm, just noticed it's printing new progress information now, in the format ‘=====> T12368a(optasm) 4655 of 6523’. <nckx>So we either add that format to Guix's sniffer and have a progress bar that goes to 100% slowly, then back to 0, then back to 100 over x hours (yuck) or add some kind of (progress-format . 'ghc) property (yuck) or… any other ideas welcome. <nckx>s/slowly/quickly/, I'm broken today. <bdju>if some accompanying text changed each time the progress started over, that sounds fine <bdju>something to show it's a different step <ng0>sneek: later tell civodul: can you moderate Jean Louis on the list? It's already clear Jean won't stop on their own. <davexunit>most of the gnu lists I subscribe to are for guile and guix, and they've sent probably dozens of emails at this point across many of them. <ng0>normally I'm okay with whatever and keep my archive as it is, but.. yikes. I'll add this adress to /dev/null <ng0>the first real human I add to spam <bandali>i think writing mostly the same email but address to different people as separate messages was a bit much <bandali>ideally they’d send one or two messages, and cc everyone they want <davexunit>they also have a page on their website that I won't link to that says really offensive stuff about ludo and others. <nckx>Flagging these as spam if you use $centralised_mail is a good deed. <nckx>davexunit: Can you tell whether it's ‘automatic’ vs. ‘reported’? <davexunit>just that gmail flagged it without me having to do anything. <bandali>i wonder if the whole ‘hatred’ thing didn’t strike them as an irony, given their own messages <ng0>this is a rather common tactic of a subset of a particular political stream, but I know too little about this person and I already invested 3 more minutes than I wanted to in this. <nckx>I mean, I'm the first to believe that even the most basic filter will pick out half of those mails as spam. <davexunit>if this were a list I adminstered I would put a ban in place. <ng0>depending on what the website says you could also file for real defamation or whatever.. <davexunit>it's a rambling screed. nothing very coherent but definitely offensive on many levels. <ng0>that is, the individuals getting insulted there. <davexunit>they also pull in many direct quotes from a 4chan thread about the topic <davexunit>which, of course, contain transphobic, homophobic, racist, etc. comments <ng0>suchserious, many wows <ng0>not surprised to say the least, given the vocabularly is very close to what polical right here use <ng0>but hey, let's not waste more time and energy on this person. <davexunit>I think ludo was too kind in responding at all. <leoprikler>I like how they repeatedly use a specific term as if they had just learned about it that day. <nckx>And, especially ironic considering the ridiculous accusations slung at him, too democratic. And that takes time. <davexunit>if I may add just one more comment. their web page has a paragraph that says something to the effect of "you can hack on an OS but you can't figure out how to enable comments on your blog." as the author of the tool used to generate the guix website I just found that very funny. <rushsteve1>davexunit: Our of curiosity how is the Guix website generated? I've been considering trying to contribute a search box for the packages page. <davexunit>rushsteve1: it uses a static site generator, which is why there is no search function on the packages page. I haven't kept up with the discussion but I think the plan was to eventually create a real web application for things like the package list so there's a true server component that can process things like searches. <nckx>leoprikler: Which term is that? ‘Rollback’? I lol'd at that. <davexunit>which reminds me that I have a pile of patches that people sent me months ago that I haven't gotten around to dealing with yet because I haven't done any free software hacking in many months. <leoprikler>nckx: wait, now I need to check whether they even mention rollbacks <ng0>hm. reminds me I should merge haunt and get back to you if there was any potential weird error going on. so this should land in the next quarterly release of pkgsrc. <rushsteve1>davexunit: Ah. If it's just a static site generator then it's probably best for me to just leave it. <nckx>It's the central theme to one of the domains. There are just too many. <davexunit>on another topic, thanks to whoever fixed the coin3d package! I can install freecad now, but unfortunately it seems to have some non-fatal error that causes the application to not be able to do much of anything. <nckx>davexunit: You're welcome and damn. <nckx>I don't actually use FreeCAD myself. <nckx>leoprikler: Back to the Ungrammatical? <pinoaffe>Hi guix~ how would I go about "locking" my X session using GDM? <pinoaffe>I'm using i3wm and know I can use i3lock, slock, xlock and the like but would prefer to lock using the display manager <efraim>I got a hash mismatch for the latest guix snapshot <davexunit>rushsteve1: I would bet that writing a small guile web server that could handle package search queries would go a long way towards solving the issue. I don't know if this has happened yet, but guix is supposed to be getting its own dedicated web server which will unblock this project. <davexunit>it's impossible to host anything other than static content on gnu.org's server. <nckx>leoprikler: I don't want to paste it here, at the expense of sane conversation. Sorry. <davexunit>that web page can be found if you really want to see it. it was directly linked to in the initial barrage of emails. <davexunit>but spreading it around further via direct linking is not a great idea. <leoprikler>davexunit: I thought it was that one. I accidentally dug it up earlier today <bdju>weston still fails to build (already reported earlier, just thought I'd mention it) <leoprikler>That's the inverse of their question. They want to lock some non-GNOME session via GDM <pinoaffe>rushsteve1: I didn't get any of the proposed solutions to work <nckx>bdju: Thanks. Do you know if there's been a bug report? (Not a Wayland user m'self.) <rushsteve1>pinoaffe: Huh, strange. Do you have gnome-screensaver installed? <pinoaffe>rushsteve1: there is no such package in guix <pinoaffe>and if I'm not mistaken, recent versions of gdm should handle screenlocking by themselves <rushsteve1>Huh. Then I have no clue. Sorry I couldn't help. <leoprikler>Have you tried locking through org.freedesktop.ScreenSaver? <rushsteve1>It says to report the bug but I want to make sure I'm not doing something wrong first <nckx>rushsteve1: Send mail to bug-guix@gnu.org. <nckx>rushsteve1: Do you use any channels other than Guix itself? <nckx>rushsteve1: If it still happens when you comment it out it's definitely bug-guix material. <nckx>Actually, it's bug-guix material in both cases, but there was some known bug with channels very recently… <pinoaffe>also: wow, that jean louis bloke is annoying <rushsteve1>nckx: I'm going to try a purge and re-install and if it sticks around I'll file a bug <nckx>jlicht, pinoaffe: Yes, my policy is to ignore them, but that only works if nobody bites *and* they have some self-awareness. It doesn't seem to work. <nckx>rushsteve1: Dunno what ‘purge’ means here but sounds good! <jlicht>bah, I am also running into some issues wrt the guix pull + channels things :/ <jlicht>and for some reason my gdm-based login no longer seems to work <leoprikler>pinoaffe: Like the gnome dbus example, but with "gnome" replaced by "freedesktop" <rushsteve1>nckx: It's an install on top of OpenSUSE. I'm gonna uninstall and rm -rf the store and related stuff and see if a clean install works. <nckx>I use 5 kustom channels and I don't :-/ <rushsteve1>I do know I had some broken packages in that custom channel so that probably didn't help <leoprikler>it's not guaranteed to work though -- on my machine it tells me it's not implemented <pinoaffe>leoprikler: ah okay, I again get no terminal output and nothing happens *nckx hopes civodul comes on-line soon. <pinoaffe>there is an org.gnome.dfeet one, but no org.gnome one <davexunit>oh man that page has even more stuff added to it since yesterday <cbaines>rushsteve1, I'm also seeing the same guix pull issue <jlicht>I am now trying without any (extra) channels <davexunit>it's honestly kind of impressive how insane it is. <jlicht>same error with only the default guix channel :/ <nckx>Damn maintainer democracy. <jlicht>davexunit: we live in the darkest timeline :P <leoprikler>"org.gnome.dfeet" should not show up when you search for ScreenSaver <nckx>davexunit: And insane bullshit is stranger than either. <davexunit>this person is a few steps away from being the next US president <davexunit>(if my country were to relax that whole "you have to be born here" law thing) *nckx was always told ‘you could be president some day’ while growing up. Now I'd be insulted. <cbaines>So guix pull --commit=0091853199bc31775c66c638a0b5b06ef0d7aec3 looks to work <pinoaffe>leoprikler: I got nothing searching ScreenSaver, so I searched for gnome instead <nckx>Ah facepalms, I just realised something. I ‘pull’ from file://~/guix… 🤦 <davexunit>yeah I think the title of President or Prime Minister has been seriously degraded in recent years <davexunit>jlicht: the absolute worst timeline where everything is darkly hilarious but there's nothing to balance it with <rushsteve1>Ok yeah after a full re-install I'm getting the same error. So I'm outputting it to a log file now and I'll email the bug list <nckx>rushsteve1: Thanks. I've rebased my ~/guix, let's see what splodes. <pinoaffe>though loginctl list-sessions does list my session <rushsteve1>nckx: Writing the bug report now, anything specific I should add? <efraim>I think for loginctl lock-session to work you'd have to build your WM with elogind <pinoaffe>I'll ask the i3 people next, seems like this isn't really guix-specific <rushsteve1>pinoaffe: Maybe try one of the GNOME specific channels too. They're the most likely to know the inner workings of GDM. <nckx>Did you try removing your custom channel? I don't see it in the backlog. <cbaines>I can't build the guix package from a checkout of the master branch <nckx>guix pull: error: You found a bug: the program '/gnu/store/fy8kjc7kymh8r1g9sb3h4ky7z8s8b7zf-compute-guix-derivation' failed 🙂 <vagrantc>i was having troubles building from a checkout before that... *nckx reverts 6c50e1dc0625f89884cff40b22627091efa37708 and pulls again. <nckx>vagrantc: I have no problems using pre-inst-env on master, but guix pull from a local repository fails. <jlicht>I worked around my issues by going back to trusty 'ol slim again, which works! My first few keystrokes into the username field of slim feel kind of sluggish though <pinoaffe>rushsteve1, leoprikler: will do, thanks for all the help! <vagrantc>nckx: i don't quite recall either... i was kind of obsessively building core-updates and merging master into it regularly the last weeks :) <nckx>Does anyone know why guix was bumped to revision 7? It's not usually done for looks. <nckx>I don't find any mentions in my mail box. <cbaines>This is where commit messages including the motivations are useful *vagrantc has often found guix commit messages a bit sparse <leoprikler>Shouldn't Guix commits always be written in ChangeLog style? <cbaines>I wonder if this is something to do with Savannah <cbaines>After reverting the changes to revision 7 of the guix package, I get the same error for a different commit error: Server does not allow request for unadvertised object 0ed97e69805253656df929a6ad678016aa81f08a <cbaines>I think this is something that's configurable with a Git server, you can configure what commits you can fetch, so maybe that's changed? <cbaines>We could test this by pushing a branch to Savannah for one of these commits, if it can then be fetched, then we've at least worked around the issue <rushsteve1>Bug report email sent. Has the stderr log attached <nckx>cbaines: That's not a fatal error and happens quite often. Are you saying it's the fatal cause here? <vagrantc>the git it was updated to happened to be the guix-jupyter commit <nckx>cbaines: I completely agree that this would be a legitimate use of a splainy commit message. <nckx>vagrantc: Ah, maybe Ludo was a bit too excited ;-) <vagrantc>fc1fe722a05318ac05a71a0b127f231631e2843f gnu: Add Guix-Jupyter. <rushsteve1>Yeah that's the commit hash that is failing for me too <cbaines>Maybe the hash is wrong, I see @ hash-mismatch /gnu/store/sin7s2f4qw3f17fs8gfv4n059dciml9j-guix-1.0.1-7.fc1fe72-checkout r:sha256 1j2d9anxgybv86pxcn1zdv121hb4nmjjp5ngx365fnd0mcg8q1iw 0yx19hxvmkr6ar65ym87xd83n6hz02mr7ibwis7i8wah85ypfq26 <rushsteve1>I just realized I put the wrong one in my email... <nckx>cbaines: I get the same actual hash. <cbaines>Right, ok, so ignore my attempt at adding a branch <cbaines>The hash for the checkout of guix at fc1fe72 doesn't match Savannah <nckx>I'm a bit busy and since this seems to be in competent hands, I'll leave you to it. <cbaines>If you get the same hash as I do, I'll go ahead and push that change <cbaines>I'm not sure quite how this package definition is breaking git pull for subsequent revisions, but it seems to be <rushsteve1>cbaines: what hash should I get and how do I check it? <cbaines>rushsteve1, if you run guix pull and get an error, look up through the output to where it says r:sha256 <cbaines>After that there should be two hashes, the expected hash, and the actual hash <jlicht>"I have no intention to disturb Guix mailing list." ... I don't even <cbaines>Looks like mbakke has pushed a fix now, so guix pull should now be back working <cbaines>It might not work for revisions between the update to the guix package, and before the fix to the hash though <rushsteve1>yeah I was getting the same hash mismatch I think <rushsteve1>@ hash-mismatch /gnu/store/sin7s2f4qw3f17fs8gfv4n059dciml9j-guix-1.0.1-7.fc1fe72-checkout r:sha256 1j2d9anxgybv86pxcn1zdv121hb4nmjjp5ngx365fnd0mcg8q1iw 0yx19hxvmkr6ar65ym87xd83n6hz02mr7ibwis7i8wah85ypfq26 <rushsteve1>Yup just confirmed it's fixed. That was lightning fast, thanks everyone! <c0c0>just wondering: why do some guix variables start with a percent sign? <bavier>c0c0: "global"-like variables or parameters, stylistic thing <ATuin>i'm getting "guile: warning: failed to install locale" everytime i run guix <ATuin>but i have the locales in /run/current-system <ATuin>any idea what am i doing wrong? <leoprikler>It might have something to do with the recent libc update <leoprikler>does it show up if you do something simple like "guix package -s guix"? <ATuin>seems guix uses guile liked against livc 2.29 <ATuin>i guess running guix system reconfigure should fix it <ATuin>i think i did 2 days ago, don't really remember <ATuin>but probably i run pull after that :) <leoprikler>perhaps a restart is required afterwards as well <leoprikler>2 days feels to old for the new libc -- you should definitely reconfigure <ATuin>ahh it's building a bunch of stuff <ATuin>ok, one more question, i have cloned the git repo to play with it but seems that everytime i pull it (specially after a while) i need to recompile it from scratch <ATuin>is that normal or am i missing something? <apteryx>do you mean because of ABI breakage (there's a warning about it when it happens) -- typically I run 'make clean-go' and rebuild everything to make sure, but supposedly you could find the affected files and purge just these. <ATuin>I don't really think it's because ABI breakage. I have the feeling that i'm doing something wrong <apteryx>OK, then what makes you think 'you need to recompile it from scratch'? do you just run 'make', and it builds lots of stuff? <ATuin>so i need to cleanup, create an env and bootstrap from scratch <ATuin>first thing i do wrong i guess it's that sometimes the environment is garbage collected so the guix wrapper points to a profile that does not exist <ATuin>i tried using a profile just for that but i could not make it work <apteryx>I also sometimes have to re-run ./bootstrap after running 'guix gc' and GC collecting older Guile, etc. <ATuin>ok, so it's normal then. thanks <apteryx>I re-run ./configure --localstatedir=/var too when this occurs. <ATuin>and for editing that source git with emacs, what approach do you use? <ATuin>i basically set the binary (pointing to the environemnt) for guile and the guile-path for geiser and i use geiser then <ATuin>dunno if there is a better approach <apteryx>it uses the guile from my main user profile <ATuin>i guess "guix-directory" and "guix-guile-program" are for emacs-guix <ATuin>btw leoprikler reconfigure worked without restarting :) <ATuin>i should have guessed that solution before asking :) <honkitybonk>My init is getting stuck on populating /etc/. I can force it to stop and it'll finish init just fine <honkitybonk>it never actually finishes if I don't force it to stop though <leoprikler>can you find out where exactly it hangs by looking at the processes? <leoprikler>I'm not quite sure what tools are available in the installation image, sadly. <honkitybonk>yeah I'm not sure what to do either since it happens before I can login and I 
