IRC channel logs
2023-03-08.log
back to list of logs
<agnem>Also, does anyone already have an implementation for a service for the interception-tools package? <fruit-loops>Exception: #<&compound-exception components: (#<&error> #<&origin origin: "scm_from_utf8_stringn"> #<&message message: "input locale conversion error"> #<&irritants irritants: 0> #<&exception-with-kind-and-args kind: decoding-error args: ("scm_from_utf8_stringn" "input locale conversion error" 0 #vu8(31 139 8 0 0 0 0 0 4 3 236 189 109 111 227 76 142 40 250 125 128 253 5 231 75 128 197 98 207 197 129 157 56 111 157 204 167 147 118 146 78 102 14 <fruit-loops>8 44 219 234 216 150 91 178 243 210 184 63 254 146 245 34 149 100 21 73 201 57 7 247 92 236 3 60 221 233 136 100 177 88 44 22 139 197 98 101 155 81 182 142 215 155 117 244 231 189 127 202 156 127 252 183 127 252 117 179 26 7 235 120 57 221 43 126 159 237 77 210 100 177 247 175 179 245 122 245 231 253 253 222 113 175 123 118 208 237 157 126 234 246 142 122 255 218 237 118 247 246 246 14 186 7 255 242 33 164 122 7 72 234 79 69 235 255 197 98 139 <fruit-loops>138 31 40 197 73 180 14 103 104 88 86 193 122 182 247 127 239 79 151 155 253 108 157 164 209 126 48 94 244 78 210 240 247 73 239 120 22 188 252 122 95 188 30 188 28 28 28 157 157 255 90 244 150 71 191 95 59 211 89 216 57 239 30 118 79 58 163 77 60 31 71 41 26 157 63 253 79 199 68 165 157 108 29 164 235 104 188 183 19 93 135 226 127 89 32 189 62 252 215 220 249 255 192 106 232 159 59 111 171 201 225 193 244 125 244 251 231 249 233 226 248 247 2 <fruit-loops>31 147 113 239 125 22 56 115 39 75 195 46 76 146 238 219 239 102 211 104 167 38 254 15 155 81 126 65 31 140 199 135 135 159 222 122 203 197 225 233 244 228 103 239 215 113 118 242 242 124 16 63 159 253 238 189 62 47 23 142 160 215 81 182 206 54 241 58 50 226 238 142 211 23 153 197 218 177 145 255 195 132 237 46 46 31 226 59 126 188 27 234 215 135 201 50 252 121 254 182 88 189 254 202 130 95 243 240 121 124 126 120 184 200 126 62 159 44 127 125 <fruit-loops>126 234 180 87 136 93 91 113 53 226 50 121 93 206 147 96 140 107 112 173 43 191 191 12 210 253 249 239 120 213 98 69 198 5 89 214 192 239 108 61 222 223 81 209 27 182 182 147 5 107 216 214 174 35 134 205 253 19 44 186 123 91 78 207 222 222 209 95 227 207 176 215 42 255 119 240 121 63 219 59 56 248 243 193 193 222 63 202 95 240 95 255 9 255 171 189 89 153 100 157 70 238 237 157 244 62 41 250 173 72 110 47 47 123 123 189 156 225 86 36 9 46 79 14 <bjc>is it supposed to be sending crash logs to the channel? <oriansj>pasting the bug to a pastebin and sharing the link here however might be valid. <lechner>that's error from the bot. the log is above it <oriansj>sorry, i was wrong, it appears to be the decimal values of ascii chars <lechner>mirai / thanks for the review. i did everything and only struggle with the activation comment. going into a school board meeting now. will get back to you. <bjc>mirai: i think what's needed is less a formal structure with file system ordering, but the ability to modify shepherd's dependency graph more intimately than is currently possible (at least wrt to file systems) <bjc>the default setup is, imho, fine, as long as i can craft a config that says "but for this file system *also* have a dependency on this other shepherd service" <bjc>that way i can say "/data is mounted over nfs, and has a dep on / and networking" <bjc>or, "/home requires zfs-import" (why i'm interested in the topic) <bjc>a "networked-file-systems" service may be useful as a generic catch point, but it shouldn't preclude other options, and, ideally, be based on something far more flexible <mirai>bjc: yeah, I think "networked-file-systems" should have a similar status to 'networking <mirai>assuming __whatever issue number was assigned to the networking canonical name patch__ <mirai>networked-file-systems is just a 'virtual name' <bjc>i took a stab at adding a "needed-by" field to shepherd, where a service could inject itself in the dependency graph. i had a minimal version working before ludo switched everything to actors <mirai>or merely provided for convenience <bjc>so i could declare a shepherd service ‘zfs-import’ and set its ‘needed-by’ to ‘file-systems’ to make sure it was done before that target was reached <mirai>the current way of things forces a very "forward view" of the graph <bjc>shepherd is, in general, pretty rigid in its configuration, and could do with letting its hair down a bit <bjc>another option would be just opening up the ‘file-system’ struct to declare deps by shepherd-service, rather than just by mount point name <bjc>that's probably(?) the easient thing. i haven't looked into it, though <mirai>that's one of the points I'm going to push for in that draft <mirai>it's sounds doable to me too <mirai>the trickier part might be loops <bjc>as long as ‘/’ and ‘/gnu/store’ are the same file system, i think it's ok. and i /think/ they have to be <bjc>worst case you can stick the networking utilities on the initrd. they may already be there anyway. if they're not, it shouldn't be too hard to put them there from the closure of stuff ‘needed-for-boot? #t’ <apteryx>lechner: thanks for the shim on debian pointers; it still escape me if the binary is signed with microsoft's key and by which organization (or by a debian developer themselve?) <winter>is there a reason Guix's GCC wrapper seemingly doesn't enable e.g. stack protection? <winter>plus various other things like PIE <lechner>apteryx / yeah, i will never trust anything microsham does, but i think they do somehow sign the shim. i would be willing to experiment with Secure Boot locally. maybe that means we all use our individualized keys, and some small part, such as 'my-shim' would always be built locally and never be delivered as a substitute <bumble[m]>for some reason, when using this guix machine, cloudfront-protected sites always block my qutebrowser <bumble[m]>the same sites do not block arch linux qutebrowser from the same network <davidl>Hi, Im trying to use emacs-webkit and it's almost working just that youtube videos hangs when Im trying to play them. I have emacs-pgtk-native-comp installed from the flat channel, and I installed emacs-webkit from source (compiled with webkitgtk-with-libsoup2) and a bunch of gstreamer and gst-plugins but it's not solving everything - tiktok works but not youtube, anyone who has any idea of what package I might be missing for youtube to work <pjals>some instances i know are inv.vern.cc yewtu.be and invidious.tiekoetter.com <wdkrnls>davidl: I have noticed that that both the webkit browsers I tried (surf, epiphany) cannot play YouTube. <wdkrnls>Heck, I can't get ungoogled-chromium to play YouTube with sound either. <Guest39>hi guix! I would like to add unicon package and rebuild noweb (already in gnu/packages) with it. what's the proper way to proceed? <AwesomeAdam54321>Guest39: You can read through the "Contributing" section of the Guix manual and modify your own copy of the Guix repository <Guest39>AwesomeAdam54321 I did.. so do I post the patch to the mailing list or create a pull request? <tux_life>Hi! I have no idea how to use "guix home". I found this other solution to automate the installation of dotfiles on the home of new users: Through "skeletons" each user gets a script to clone my Github repository with all the dotfiles. What do you think of this solution? <evilsetg[m]>I am trying to install guix with lvm on luks. I made it into the initrd after adding an 'insmod lvm' into the grub.cfg. Now I am dropped into a guile repl after the message 'pre-mount actions failed'. Unfortunately, ',bt' does not give me much. It tells me that the error occurred in (gnu build linux-boot). Can anybody help me in diagnosing where this boot fails? <TristanCottam[m]>Hi guys, does anyone know what it takes to add packages to a desktop environment? I'm wondering if`seahorse` (GNOME's "Passwords and Keys") could be included in GNOME's dependencies upstream. <TristanCottam[m]>tux_life: I'd recommend taking the time to learn about Guix Home before considering alternatives, as it fits in with the rest of the GNU system and philosophy <tux_life>TristanCottam[m] Thank you! I understand... But for the moment it's not even clear to me what is the result that can be reached via Guix home. I don't know if I should specify the content of each individual dotfile, which would be a lot of work, and how. <TristanCottam[m]>You could do virtually anything you could think of, including linking external files inside you config <TristanCottam[m]>The real work will be learning about Guix in general, but if you're ready to put in the effort, it's well worth the effort <TristanCottam[m]> * The real work will be learning about Guix in general, but if you're ready to put in the effort, it'll pay off in time <tux_life>TristanCottam[m] This is true! And I'm trying to learn...just because I see the benefits. :] <oleander>Hello, I'm getting a weird behavior with Guix when opening an html file generated with the makeinfo command. I'm using this css file https://codeberg.org/luis-felipe/texinfo-css to generate the html file specifying the --css-include option. Now every browser I tried in Guix does not display the css properly. Colour background for code blocks is missing for example. I don't think this has to do with Icecat or Ungoogled Chromium, <oleander>since the same file synced with my smartphone is not displayed correctly in chromium. I emailed Luis and he confirmed the same behavior in Guix where Icecat and U.Chromium don't apply the correct color scheme, whereas with Epiphany everything is displayed correctly (in my case it is not). If I instead generate the html file running the same makeinfo command in Termux I don't get this behavior. Is there something that I'm missing <gabber>i get `Could not find the Qt platform plugin "wayland" in ""` with several programs since i've switched to Wayland (sway); e.g. flameshot and wireguard. am i missing some package? installing `qtwayland` did not help. <davidl>wdkrnls: thanks for checking. Oh weird with chromium. I wonder if it's a webkit regression then. <davidl>AwesomeAdam54321: it works! Thank you so much!! <ellysone[m]><gabber> "i get `Could not find the Qt..." <- For flameshot I had to make variant <ellysone[m]>IIRC there is an issue about adding qtwayland by default so that you don't have to define variants like that but I can't find it rn <ellysone[m]>in the meantime you can do that, or export QT_QPA_PLATFORM=xcb <gabber>ellysone[m]: thanks! why didn't you opt for the (seemingly easier) set the env-var hack? <ellysone[m]>gabber: IIRC the xcb with flameshot wasn't working properly for me, like the cursor <gabber>btw the env-var also fixes the error message with wireshark <gabber>yeah, you're right. flameshot isn't working properly (yet) <dcunit3d>how do i remove pulseaudio from my system if i'm on X11/i3? <dcunit3d>i've tried removing pulseaudio-service-type from my operating system services and i tried switching back to slim from gdm <dcunit3d>shepherd doesn't list it in services, but it gets restarted after i kill the process <dcunit3d>i may convert to sway anyways. i'm almost ready to do that, but not quite. <bjc>i think pulseaudio gets started from dbus <dcunit3d>that i could switch so easily from GDM to slim is pretty miraculous lol <gabber>i'm back with my question :) what's the easiest way to define a one-shot service which executes a command? e.g. i need to add an address to an ethernet interface and i need to write "1" to /proc/sys/net/ipv4/ip_forward <ellysone[m]><gabber> "i'm back with my question..." <- you can handle the ip_foward with the sysctl-service and as for one-off service I don't have anything on the top off my head, best is to check the source code of guix/gnu/services <mirai>gabber: have you tried using the native facilities of the network manager you're using <dcunit3d>ellysone[m]: how is it built? i don't see much on the readme and there are also not any files that list dependencies <ellysone[m]>dcunit3d: you're supposed to use go generate which will generate metro_amd64.s and from there go test <gabber>mirai: i'm not sure whether (or how) that would be possible since i need wpa-supplicant for a dhcp-connection on one interface and i NAT connections with dnsmasq on another interface (which i need to assign a static IP to) <mirai>are you using connman or networkmanager? <dcunit3d>i would help you if i could, but i have minimal experience with go. is the parent project written in golang? <ellysone[m]>in go-hasher all the code related to go-metro is commented out <gabber>mirai: (i think) neither -- they would show up in `sudo herd status`, no? there's only wpa-supplicant (which i need to get a dhcp lease from our network) <mirai>they show up as 'networking' <mirai>wpa-supplicant doesn't do dhcp <dcunit3d>ellysone[m]: you could try generating a patch that comments it out from go.mod and see if it builds. if go-metro isn't also depedency of the other packages in go.mod, then it might work <gabber>mirai: i see. checking %desktop-services i think it's network-manager <mirai>check NetworkManager documentation <ellysone[m]>well yeah, but I do want to know how you're supposed to build that package from source in 2023 with the code that's in the repo <mirai>it has some knobs you can turn and it can also run scripts upon certain criteria <gabber>mirai: ok, cool, will do! thanks! <dcunit3d>ellysone[m]: if you look into ./guix/build/go-build-system.scm and ./guix-build-system/go.scm, you may be able to add a phase before the "build" phase where "go install" runs <TristanCottam[m]>Hi again, everyone! Does anyone know if it's possible to submit a patch to Guix based on the contents of a custom channel? <gabber>mirai: how can i integrate NetworkManager's configuration with my system configuration? <TristanCottam[m]>I want to maintain my own channel, but would like to be able to easily submit some of my code upstream <gabber>TristanCottam[m]: you can write your own packages and maintain your own channel but you will most certainly have to edit your patches when submitting them upstream <TristanCottam[m]>Of course, but is there a "seamless" way to convert the changes from one to the other, or can it only be done manually? <gabber>TristanCottam[m]: by definition: no, since you will have to create your own module structure. but you can imitate the upstream structure as closely as possible. usually not too many changes are necessary <dcunit3d>TristanCottam[m]: the guys at the https://sr.ht/~whereiseveryone/guixrus channel might also be able to help you out. they're at #whereiseveryone on IRC. if your changes are larger in scope, then it may be easier to fit it in there first with the intent to merge it upstream later. <TristanCottam[m]>I'll check it out, thanks! Do you have experience in building packages for Guix? <dcunit3d>minimal. i've pulled in some things i need into a personal channel, but nothing that is worth merging into master. i'm hoping to get more into it soon. <dcunit3d>i would like to work on a squashfs bootloader, some PXE stuff and maybe a build system for swift (for the substrate graphics framework), but the last one is already way out of my league <gabber>i have written a couple of package definitions, yes (: why do you ask? <TristanCottam[m]>Could I DM you about a couple of questions? I tried here already with no luck :( <gabber>if it's nothing frowned upon i think you could also (re-)post them here <gabber>but you can also DM me, of course <cow_2001>anyone got a simple guile standalone (as in a single executable) guix package from which i could steal the skeleton for the guix package file? <gabber>cow_2001: i think you can find some in gnu/packages/guile-xyz.scm <gabber>is it possible to get some insight as to why QA failed to build my patch #61869 ? i was able to build and test it seamlessly <civodul>cow_2001: there's the 'guile-static-stripped' package <civodul>but it's not a single file and cannot be a single file currently <mirai>gabber: you can partially integrate it <mirai>.nmconnnection files you will have to manually provision them (place them under /etc/NetworkManager/....) <mirai>you'll have to check both NM manual and Guix manual to see which configuration goes where <mirai>gabber: regarding QA, click on it and it takes you to a page <mirai>occasionally QA fails due to unrelated matters so don't worry too much about it <gabber>i checked the QA page out but it didn't give me info on what step or why it failed to build (or i didn't know where to look) <mirai>:( mumi doesn't display the patch <mirai>your patch doesn't change any package definition <mirai>its some packages failing to build due to $REASONS <gabber>aaah, i see :) thanks for checking! <gabber>i just hope that doesn't delay the merge <mirai>minor nitpick, your documentation for dhcp-range could be changed to use 192.0.2.0–192.0.2.255 as IPs <attila_lendvai>civodul, i've pulled and reconfigured some new service code, and it looks to me that it didn't get actually used until a rebooted. is this a known limitation, or am i doing something wrong? <civodul>attila_lendvai: 'guix system reconfigure' won't stop and restart services that are currently running; you have to explicitly "herd restart" them to get the new service in place, or to reboot <mirai>gabber: btw, are the default values for the newly added fields the defaults for dnsmasq? <gabber>they should be. but i took some liberty in renaming some of the options (one is called --no-poll which i converted to the option poll?) <attila_lendvai>civodul, that i know, and i did. but it still seemed to run the old code. but if i look at the init process in ps, then shepherd is started with a .conf file, and that file contains the .go file of my service code, in the store. this suggest that what i see is exectable, no? <attila_lendvai>i mean, shepherd on pid 0 is not restarted unless i reboot, right? <mirai>gabber: but when it comes to renaming negated options like that be sure to check what the manual says <mirai>it's not always "binary", some programs admit a "third value" that is 'unspecified' which has different behavior <civodul>attila_lendvai: right, shepherd is not restarted until you reboot; there's no way around it <gabber>mirai: do you have anything particular in mind? i only added options that i needed to set, and i had close looks at dnsmasq's --help. but of course some things could've slipped my attention <attila_lendvai>civodul, well, theoretically we could unload and reload the .go files and the relevant modules, but i guess even if that is possible in guile, it's not a code path that has been exercised much. <mirai>I don't believe it fits within a mere dnsmasq --help <gabber>the problem with the manpages is that a `guix shell dnsmasq -- man dnsmasq` wouldn't show it to me <mirai>the page should be something ...thekelleys... <gabber>mirai: i'm getting confused. is there anything particular i should look up there? what should i look out for? <civodul>attila_lendvai: right, and it's PID 1, we'd rather be cautious :-) <mirai>gabber: just confirm that: the default values match dnsmasq defaults (try to preserve the default behavior unless you feel its justified to stray from them); when renaming negative options such as "--no-poll" confirm that: does "--poll" exist? is it really the negation of "--no-poll"? (say that "--poll" doesn't exist but you set the field to #t. Although it's an empty argument, the program behavior doesn't imply that any "polling" will be done) <gabber>i will and send in an updated patch! thanks for your time! <cel7t>civodul: when is `guix index` expected to be merged? <civodul>cel7t: there's no ETA but i believe the person who started it is still working on it <cel7t>I had a question regarding guix install/remove <cel7t>The build phase after installing/removing packages seems to be very long (it can take a few minutes after installing about 30 or more packages) <cel7t>Is there a way to shorten the build phase? <civodul>cel7t: are you talking about the build phase of a specific package? <lechner>Hi, is there a space missing after @deftp here? I get doc/guix.texi:32489: (po4a::tex) unmatched end of environment 'deftp' <civodul>lechner: i've seen that too, must be a recent regression <lechner>i'll write a patch for it if someone here commits, or someone can do it own their own <mirai>should be @deftp {Data Type} <mirai>kinda curious but I don't see the warning <cel7t>civodul: no, packages in general <cel7t>I have observed that after reaching a certain number of packages the build phase takes a while <cel7t>But it seems that I might just have forgotten to enable substitutes Σ(ﾟДﾟ； <fruit-loops>Exception: #<&compound-exception components: (#<&error> #<&origin origin: "scm_from_utf8_stringn"> #<&message message: "input locale conversion error"> #<&irritants irritants: 0> #<&exception-with-kind-and-args kind: decoding-error args: ("scm_from_utf8_stringn" "input locale conversion error" 0 #vu8(31 139 8 0 0 0 0 0 4 3 236 189 109 111 227 76 142 40 250 125 128 253 5 231 75 128 197 98 207 197 129 157 56 111 157 204 167 147 118 146 78 102 14 <fruit-loops>8 44 219 234 216 150 91 178 243 210 184 63 254 146 245 34 149 100 21 73 201 57 7 247 92 236 3 60 221 233 136 100 177 88 44 22 139 197 98 101 155 81 182 142 215 155 117 244 231 189 127 202 156 127 252 183 127 252 117 179 26 7 235 120 57 221 43 126 159 237 77 210 100 177 247 175 179 245 122 245 231 253 253 222 113 175 123 118 208 237 157 126 234 246 142 122 255 218 237 118 247 246 246 14 186 7 255 242 33 164 122 7 72 234 79 69 235 255 197 98 139 <fruit-loops>138 31 40 197 73 180 14 103 104 88 86 193 122 182 247 127 239 79 151 155 253 108 157 164 209 126 48 94 244 78 210 240 247 73 239 120 22 188 252 122 95 188 30 188 28 28 28 157 157 255 90 244 150 71 191 95 59 211 89 216 57 239 30 118 79 58 163 77 60 31 71 41 26 157 63 253 79 199 68 165 157 108 29 164 235 104 188 183 19 93 135 226 127 89 32 189 62 252 215 220 249 255 192 106 232 159 59 111 171 201 225 193 244 125 244 251 231 249 233 226 248 247 2 <fruit-loops>31 147 113 239 125 22 56 115 39 75 195 46 76 146 238 219 239 102 211 104 167 38 254 15 155 81 126 65 31 140 199 135 135 159 222 122 203 197 225 233 244 228 103 239 215 113 118 242 242 124 16 63 159 253 238 189 62 47 23 142 160 215 81 182 206 54 241 58 50 226 238 142 211 23 153 197 218 177 145 255 195 132 237 46 46 31 226 59 126 188 27 234 215 135 201 50 252 121 254 182 88 189 254 202 130 95 243 240 121 124 126 120 184 200 126 62 159 44 127 125 <fruit-loops>126 234 180 87 136 93 91 113 53 226 50 121 93 206 147 96 140 107 112 173 43 191 191 12 210 253 249 239 120 213 98 69 198 5 89 214 192 239 108 61 222 223 81 209 27 182 182 147 5 107 216 214 174 35 134 205 253 19 44 186 123 91 78 207 222 222 209 95 227 207 176 215 42 255 119 240 121 63 219 59 56 248 243 193 193 222 63 202 95 240 95 255 9 255 171 189 89 153 100 157 70 238 237 157 244 62 41 250 173 72 110 47 47 123 123 189 156 225 86 36 9 46 79 14 <fruit-loops>63 247 180 251 94 17 166 159 205 94 247 240 190 21 73 227 148 214 140 79 175 123 212 148 36 97 50 246 246 78 207 142 112 128 62 90 152 189 83 28 40 137 48 255 84 245 186 55 97 24 69 227 178 223 189 227 228 254 147 223 232 198 135 189 244 168 55 93 188 197 193 98 18 156 190 255 30 253 158 76 166 105 124 252 220 27 189 190 47 62 41 163 11 97 137 131 238 145 124 209 109 72 212 53 169 98 221 58 58 59 19 75 56 39 202 171 194 167 30 154 4 201 192 85 <fruit-loops>6 184 88 76 78 222 178 163 211 209 39 248 105 5 67 245 243 165 23 246 206 70 233 201 225 233 113 207 172 148 7 221 195 78 49 57 229 195 215 154 60 49 144 254 25 125 114 126 32 148 184 104 170 52 223 250 18 130 62 12 127 134 211 95 207 217 244 125 122 180 58 254 116 182 250 29 78 15 142 222 143 159 95 198 227 215 241 111 189 23 56 235 158 194 62 186 206 178 202 92 212 29 27 169 21 58 175 232 135 135 199 31 41 246 93 23 105 98 16 38 47 47 179 23 <fruit-loops>166 113 239 237 244 237 124 122 116 246 154 158 157 143 38 206 62 161 141 182 183 38 239 10 190 129 163 182 227 112 151 188 25 191 234 225 202 118 220 198 245 96 73 158 29 74 13 175 99 118 125 51 196 176 121 252 233 84 172 140 13 68 221 218 144 85 132 188 109 72 145 239 115 199 115 108 232 63 156 117 105 146 45 28 39 142 228 233 39 249 124 111 32 226 134 75 125 73 176 133 91 225 250 181 232 99 23 255 53 20 44 71 18 34 158 77 221 71 142 164 220 <fruit-loops> 241 29 205 25 97 247 103 207 179 116 241 58 10 95 39 191 179 44 59 60 206 142 151 179 73 112 254 41 56 254 253 190 12 142 126 23 94 142 220 181 105 70 179 165 133 111 189 162 148 102 135 14 207 151 87 52 216 80 30 232 13 165 158 33 13 103 7 71 242 252 72 106 35 114 211 206 145 60 57 150 250 211 162 217 209 90 180 148 158 77 143 199 39 163 223 207 171 131 247 244 57 27 255 58 90 126 58 94 156 30 254 6 175 250 236 253 244 117 188 163 127 49 107 <fruit-loops> 190 117 103 4 170 125 122 118 210 126 69 173 215 195 130 164 220 74 51 170 93 144 60 61 254 36 118 90 4 155 197 214 162 45 130 91 202 7 64 99 232 95 79 155 70 58 56 146 135 221 211 166 11 42 71 82 62 84 34 147 209 108 62 16 118 34 252 125 248 252 254 124 126 240 242 59 11 199 39 103 7 243 24 162 211 47 111 7 63 95 14 127 101 225 115 108 214 163 227 238 177 124 61 106 70 179 165 69 104 214 72 105 166 158 65 103 84 119 62 82 163 104 146 173 52 <lechner>sorry about that. cel7t / please repost <lechner>yeah, i see that. guix assumes there is a terminal. it's not the first time i have had that issue. i also get that when trying to read the system time from guix. i get the copyright notice! <gabber>mirai: i've sent an updated patch for #61869 <lechner>civodul / thanks for that pointer! lfam is asking me to fix the feature branch, but i think the timeout for pandoc, or it's origins in ghc or Cuirass, are outside the responsibilities for a Gocryptfs/go-update feature branch https://issues.guix.gnu.org/61989#13 <lechner>does "silent" mean no terminal output, or script did not return in #61839 <civodul>lechner: "asking me" and "outside the responsibilities" is not how i'd frame it since we're on the same boat, but yes, it's not related to the changes in that branch <lechner>civodul / sorry, i did not mean to sound standoffish. i do care about the project as a whole, as you saw from my guix.texi post. that probably was not my best choice of words <lechner>mirai / i would like to help you stabilize your connection if you have identified the source of your problems. i can provide anything from a new router, an SSH keep alive, or a Pringles container to refocus your Wifi <lechner>lfam just asked me: "For a while, setting the timeout and max-silent-time in a package definition didn't work here, because Cuirass didn't respect those options. I don't know if that's changed. Can you ask on IRC or guix-devel about this?" https://issues.guix.gnu.org/61989#15 <mirai>lechner: yeah, I'll get this sorted someday <lechner>mirai / my private window with you has thirty lines of valuable exchange, and three hundred Quit and Re-join notices from Circe! <ellysone[m]>I think I solved my problem, I was running my guix shell with the -D flag, without that flag go test just works, I don't know what's in the development input of go that causes it to fails <mirai>ugh, how can I trigger these substitute messages <fruit-loops>Exception: #<&compound-exception components: (#<&error> #<&origin origin: "scm_from_utf8_stringn"> #<&message message: "input locale conversion error"> #<&irritants irritants: 0> #<&exception-with-kind-and-args kind: decoding-error args: ("scm_from_utf8_stringn" "input locale conversion error" 0 #vu8(31 139 8 0 0 0 0 0 4 3 236 189 109 111 227 76 142 40 250 125 128 253 5 231 75 128 197 98 207 197 129 157 56 111 157 204 167 147 118 146 78 102 14 <fruit-loops>8 44 219 234 216 150 91 178 243 210 184 63 254 146 245 34 149 100 21 73 201 57 7 247 92 236 3 60 221 233 136 100 177 88 44 22 139 197 98 101 155 81 182 142 215 155 117 244 231 189 127 202 156 127 252 183 127 252 117 179 26 7 235 120 57 221 43 126 159 237 77 210 100 177 247 175 179 245 122 245 231 253 253 222 113 175 123 118 208 237 157 126 234 246 142 122 255 218 237 118 247 246 246 14 186 7 255 242 33 164 122 7 72 234 79 69 235 255 197 98 139 <fruit-loops>138 31 40 197 73 180 14 103 104 88 86 193 122 182 247 127 239 79 151 155 253 108 157 164 209 126 48 94 244 78 210 240 247 73 239 120 22 188 252 122 95 188 30 188 28 28 28 157 157 255 90 244 150 71 191 95 59 211 89 216 57 239 30 118 79 58 163 77 60 31 71 41 26 157 63 253 79 199 68 165 157 108 29 164 235 104 188 183 19 93 135 226 127 89 32 189 62 252 215 220 249 255 192 106 232 159 59 111 171 201 225 193 244 125 244 251 231 249 233 226 248 247 2 <fruit-loops>31 147 113 239 125 22 56 115 39 75 195 46 76 146 238 219 239 102 211 104 167 38 254 15 155 81 126 65 31 140 199 135 135 159 222 122 203 197 225 233 244 228 103 239 215 113 118 242 242 124 16 63 159 253 238 189 62 47 23 142 160 215 81 182 206 54 241 58 50 226 238 142 211 23 153 197 218 177 145 255 195 132 237 46 46 31 226 59 126 188 27 234 215 135 201 50 252 121 254 182 88 189 254 202 130 95 243 240 121 124 126 120 184 200 126 62 159 44 127 125 <fruit-loops>126 234 180 87 136 93 91 113 53 226 50 121 93 206 147 96 140 107 112 173 43 191 191 12 210 253 249 239 120 213 98 69 198 5 89 214 192 239 108 61 222 223 81 209 27 182 182 147 5 107 216 214 174 35 134 205 253 19 44 186 123 91 78 207 222 222 209 95 227 207 176 215 42 255 119 240 121 63 219 59 56 248 243 193 193 222 63 202 95 240 95 255 9 255 171 189 89 153 100 157 70 238 237 157 244 62 41 250 173 72 110 47 47 123 123 189 156 225 86 36 9 46 79 14 <lechner>i'll leave the bot off until further notice <lechner>it could be something about the terminal setup in Cuirass. I also recently noticed similar behavior in Guile. system* and open-pipe* inherits the parameters for the input port to the subordinate process from the input port of the caller caller. while that is common for scripting languages like bash (i think) we may need to turn it off https://www.gnu.org/software/guile/manual/html_node/Pipes.html <mirai>no, it has to do with whether the port is a tty or not <mirai>question is how to test this <lechner>you are assuming guix doesn't already do it. <lechner>then why don't you see those messages locally? <mirai>because interactive terminals understand escape sequences <lechner>any program that's been around as long as Guix should check for presence of tty. i would be surprised if your snippet isn't already in there <mirai>if you redirect the output to a file it simply receives the stream of bytes as is <lechner>with the terminal control character? <mirai>save for the bizarre case with piping binary data <mirai>try crafting a toy program that issues things like backspace and other control characters <lechner>that's not the question. the question is whether guix issues the terminal control sequence because (1) it always does so, or (2) because it assumes by accident that it is connected to a tty. i have seen the latter often in CI settings, particularly on Gitlab <mirai>that paste was originally only the display procedure call <mirai>well, thank for tests/substitute.log <jackhill>what's up with the 'I have no name!@host' prompts in `guix pack -RR` bash? I assume it's something to do with the relocatable tricks, but `whoami` and `id` still seem to know who I am. <Guest74>I installed Guix System with / and /home partitions. Now I want to merge everything back to /. Most tutorials use rsync or something else and I wonder can't I just use the mv cmd to move /home to /? <rekado>Guest74: you can do that. You can use rsync for a final pass if necessary. <Guest74>Do you mean "final pass" as verification the data is correct? <rekado>jackhill: I suppose this depends on the default prompt. <Guest74>I also wonder about Guix System. My plan is to reconfigure the system with only / and reboot to a live environment. Is that okay or would that make an issue since reconfiguring Guix System by removing /home partition while I run on it? <lechner>is there anything to compare after 'mv'? <lechner>i think the config has to match your mounts in order to boot <NewUser-Basic-Qu>hi all...question about sshd. Installed openssh-sans-x. Thought I will have the possibility to simply start it as on other linux or have system-service available. When executing sshd manually it says "sshd re-exec requires execution with an absoulte path" ... re-exec? I am so cinfused at this point <jackhill>rekado: indeed. Mine seems to be \[\e[32m\]\u@\h \[\e[34;1m\]\w $\[\e[0m\] <rekado>NewUser-Basic-Qu: I’d use the system service instead. <rekado>NewUser-Basic-Qu: but if you really must use it manually: use the absolute file name <rekado>i.e. not sshd but whatever readlink -f $(which sshd) says <NewUser-Basic-Qu>wow...if i look into folder which si returned by readlink/which-command it does not contain sshd... <bumble[m]>I noticed many packages used with sway wm include install directions for different systems and/or their README lists supported systems -- directions and lists of supported systetms always include NixOS but usually do not include GNU Guix <NewUser-Basic-Qu>Guys...just asking...no need to answer but how many here work as actual system-administrators? Guix is giving me such a hard time at the moment...Maybe for many here it is like a open book but for me it is really hard. Like something purely done for people with an purely academic background <lechner>NewUser-Basic-Qu / i have used Linux for thirty years. When i switched to Guix a year ago, the first month was one of the most difficult times in my life. The good news is that a static config more or less makes traditional sysadmin positions obsolete. <lechner>i have also never had such a stable system <NewUser-Basic-Qu>bumble[m] willread . lechner, i do not think so regardinf trad. sysadmin positions. At the moment i would panic using Guix in any commercial setup as i would be afraid of not being able to hire manpower to do analytic work or administration when neccessary. <lechner>you'd be surprised of the talent lurking in the shadows here <NewUser-Basic-Qu>let alone that i would not be able to check what my team actually does when they call it "work"... <lechner>for us, work is a function like this work (time) -> money <NewUser-Basic-Qu>lechner true regarding talent but in this case i am not sure where i would find the Guys/Girls. <lechner>--- Let's try it. Anyone looking for Guix sysadmin job? Please speak up! --- <mirai>I don't think it's exactly "rocket science" <lechner>the mailing list guix-devel may be better, but people ask here from time to time <lechner>NewUser-Basic-Qu / mirai is a rocket scientist <mirai>surely there are people here who began with zero knowledge of scheme and picked it up while tinkering with guix <mirai>"it's just some parentheses" <mirai>imo it's doable to pick the required skills on-the-go <mirai>even faster if you have any prior experience or "tutor" <mirai>but overall it can be self-taught (provided you ask for help here when stuck) <mirai>lock someone in a room with water, bread, computer and internet connection and they'll figure it out within a quarter <NewUser-Basic-Qu>just as an example. One of customers has a cascading data-sync-chain provided by it's own sottware(a catastropy) which we run on Head/Main -> intermediate-Servers->8000 edge-Devices and back. All running Debian. The edge Devices are running on customer side and about 200 of them a month need some form of administrative tasks or analysis, mostly <NewUser-Basic-Qu>network related. As it is how it is we have to do a lot of it via manpower instead of config-management. We hope to change this when customer presents new software and also new hardware for the edge-entities. We wondered if NixOS or Guix could be the way to go. This is why i am here... <jonsger>NewUser-Basic-Qu: whats the disk size of the edge devices? <NewUser-Basic-Qu>jonsger currently 2TB as we broke up the raid as customer could not do houesekeeping of data synched before we ran out of space. Future with new appliances points to 4-8TB and more...depending on what the nedpoint requires <lechner>NewUser-Basic-Qu / our position is that Guix will eventually replace deployment tools like Ansible and Puppet. In fact, I only use 'guix deploy' <NewUser-Basic-Qu>lechner situation on our side is catastrophic. Framework for using ansible effectively and in coordination with customer requirements(safety concerns as we access their internal networks) was developed and then never saw the day of light again. Puppet is nice ...would have been nice throughout the whole project and desperately needed often times <NewUser-Basic-Qu>but also all efforts to get it rolled out met concerns(i assume) and also was not used ..ever...crazy... <lechner>NewUser-Basic-Qu / If you really have that kind of an opportunity, I'd be happy to help you draft an email to our mailing list. I think you would attract a lot of competent interest. Folks here manage many hundreds of machines, mostly for science https://hpc.guix.info/about/ <NewUser-Basic-Qu>lechner We will need people in position of administrators. We either get the current staff to go through what i go through or we have to get rid of them(almost impossible) and replace them but people with this skillset for the price are almost impossible to find... <lechner>Well, you are in the right place. Many people here were Nix contributors before Guix got going. We are a little bit like a second-generation Nix. People will take hourly side jobs just to help you out. We all believe Guix is the safest, best, and most rational OS around. Maybe one day you will see it, too. There are plenty of docs inthe cloud <NewUser-Basic-Qu>lechner i already think Guix is the right way conceptually. NixOs was brought up by one of the Project Responsibles from Customer side. My hope is they step back from dictating the Host-system setup at the moment <lechner>just have them look at the configuration language. <lechner>there isn't an IT manager in the world that will dislike ours <mirai>lechner: have you heard of DSSSL and XSLT? <lechner>NewUser-Basic-Qu / it's so easy, an intern can change it. if it does not work, you simply boot a previous generation. there is no cost to messing it up <NewUser-Basic-Qu>I don't hink they ever considered that. The base thought on their side might be that they want an immutable system of some form as a base ... This however can be achieved in a lot of ways and with other competing linux-distros. However...It was not silverblue or any of the newer derivates that was mentioned but it was Nix. We have some time until <lechner>i'll go with you. where are you located? <lechner>bummer. i am german but live in California <mirai>bumble[m]: have you checked out the latest commit on that repo? <lechner>you sound great. where are you from? <mirai>it explains why your suggestion was discarded the way it was <NewUser-Basic-Qu>Was born and raised in Germany but still hold the foreign passport ... have to apply at some point for citicenship... <lechner>my whole family is international. my wife is from Beijing <apteryx>lechner: try jami instead, it's free software :-) <lechner>yeah, i used it when it was still called SFLPhone but then worked with the Linphone folks for a little while <NewUser-Basic-Qu>Also ich werd hier noch oefters sein. Muss erst mal grundsaetzlich mit dem ganzen Konzepten vertraut werden. Bin simpler sys-admin und das problem ist einfach wie erhalten wir die Abteilung ohne zu grosse aenderungen und koennen wir im Notfall was von den Kollegen abfangen... <lechner>NewUser-Basic-Qu / Bitte nimm Dir soviel Zeit wie Du brauchst. Wir sind auch immer zu Fragen offen. Guix - you will love it™ <NewUser-Basic-Qu>Thank you man. Currently trying to get familiar with it by trying to do what i would do on other systems to when starting to use them. See how i can start/stop, see how i can adapt workspace a.s.o. ... my arrogance expecting that everything should easily be solvable because it was with standard-linux distros ... <lechner>NewUser-Basic-Qu / i will explain everything to you. which distro are you coming from, please? <lechner>Okay, we have a Gentoo emulation mode. that's when our substitute server fail, and you have to build everything at home <bumble[m]>mirai oh no the project is being discontinued! afaik this is the only common backlight utility that works w/out dbus <NewUser-Basic-Qu>lechner the "Gentoo emulation mode" ? ...it did...Im sure you guys don't call it that way. <lechner>NewUser-Basic-Qu / anyway, Guix is very different in that we attempt to reference all executables and other references via absolute paths into /gnu/store. For ELF, we modify RPATH or RUNPATH. For scripts, we use environment variables. When all fails, we construct a fake filesystem from symbolic links that looks like what you are used to. We call the latter a "profile" <lechner>the advantage is a complete isolation (with respect to prerequisites) among all executables. different users can install and run any version of any software they like, without conflicts or superuser approval of any kind <lechner>that is more or less also true for Nix, I believe <lechner>you can easily see what i mean by booting one of our live systems and looking at /bin. It is empty (except a 'sh' in some cases) <lechner>you may not need that by-user flexibility, but what it also does is provide to each program exactly the prerequisites it was built with. we rebuilt our "archive" all the time <lechner>Debian, by comparison, never rebuilds anything. instead it relies on maintainers uploading ABI compatible versions of shared libraries. It also employs several safeguards, like dh_shlibdeps to track ELF object symbols. Unfortuately, it's never 100% reliable. In Guix, that problem does not exist <lechner>i used Debian since its inception in 1995 or so and have never had a stabler system until i switchted to Guix. Especially X upgrades were problematic for me in Debian <lechner>the absolute path thing is initially mind-boggling, though, so that why i pointed it out. once you get past that---and for you it may mean to shed decades of preconceptions---you may see that is simply the best way to do things™ <NewUser-Basic-Qu>Oh, i love the complete isolation part. the by-user flexibility would also come in handy. Doing things differently than before would be ok but i am so surprised of my inability to get a clear overview of what i have at a current moment and what not. I guess Guix repl will somehow become my best friend in near future <lechner>well, i lived and breathed Debian and I had a really hard time with Guix in the beginning. Somehow, it seemed right, yet I could not get my mind around it. Suddenly, I felt really small <lechner>Guix also rekindled my fascination with computers, however. the people here are top notch. many have academic degrees. we are friendly and cooperative. you can look around IRC for comparison <NewUser-Basic-Qu>for example the sshd situation today. I expected that by installing openssh-sans-x i will get a start/stopable service with it, that i can simply execute the command aso..having to run a extra round until it worked and not having the service available after package isntallation feels weird... <lechner>NewUser-Basic-Qu / the repl is a bit hard until you understand quoting styles in Guix. to get the full hang of it, you may need to read the Guile section about , ' and ` <NewUser-Basic-Qu>currently trying to follow andrew tropins videos on guix/scheme to get somewaht of a feeling for it <lechner>Andrew is great, and also one of Guix's most valuable contributors with respect to Guix Home <lechner>I might also recommend David Wilson's videos at systemcrafters.cc He is probably Guix's most effective spokesperson <attila_lendvai>i have two guix system installs, and one of them has the binary `locale` while the other doesn't in the path. i can't find the difference... any hints? both use guix home, and almost the same config, modulo some packages <attila_lendvai>and my root issue is that some service is logging garbage into the log file on the one that doesn't have the binary <lechner>I only see 'ssh-daemon' when i issue the same command. i must admit that i have never given it another thought before (other than looking for and publishing my SSHKEYS in DNS) <lechner>NewUser-Basic-Qu / are you using a mixed system? <lechner>attila_lendvai / is that just your reader malfunctioning? what are your personal LC_* variables both setups? <NewUser-Basic-Qu>muahaha...define "mixed system" in this context. ...btw. just.cooking a leberkaese with spaetzle...don't knowwhat the situation is with home-food in Cali for you... <lechner>NewUser-Basic-Qu / all from Sichuan for me! <vagrantc>NewUser-Basic-Qu: in guix, packages are just a bunch of files in a directory ... services typically make use of packages to provide server-like things <vagrantc>honestly, that is one of the separations i really like about guix <attila_lendvai>lechner, both of them only has LANG=en_US.utf8 in `set | grep LC`, and `set | grep LANG`. but i found a bug, i set the locale for one of the service to LC_ALL=en_US.UTF-8, which is not valid i think, it must be en_US.utf8 <lechner>attila_lendvai / can you compare the pertinent log parts with a binary tool like 'cmp'? <lechner>or look at their encoding with iconv on another machine? <NewUser-Basic-Qu>vagrantc at which point is a service kinda "available". Only if i explicitely define usage in system configuration or can i simply install the package and then invocate the service (let's say) from a different "source" ... sounds stupid...i hope that this question somehow made sense... <lechner>you have to reconfigure, and then start it <vagrantc>NewUser-Basic-Qu: typically if defined ina system configuration ... although there are also user-defined services as well ... less familiar with those ... you can of course always run things manually; nothing stops you <cel7t>Hi is packages.guix.gnu.org down? <cel7t>I'm getting a "504 Gateway Time-out" <PurpleSym>lechner: You’re Felix Lechner? The one with the Haskell rebuild issue? <cbaines>I've restarted packages.guix.gnu.org, which seems to have brought it back <PurpleSym>Okay, but which evaluation does it belong to? <lechner>the failures are hard to navigate, however, due to the pandoc failure <lechner>also, please know that i am more of a Haskell than a Golang person. i only ended up with this experimental job because I use Gocryptfs <PurpleSym>Hm, but build 478772 belongs to a build for master, not the go-team jobset. <PurpleSym>Afaik that job was never restarted after I fixed the max-silent-time property. <PurpleSym>Hm, no idea why max-silent-time is not working here 😕 <winter>is there a reason Guix's GCC wrapper, from what I can see, doesn't enable e.g. stack protection or PIE? or am i just looking in the wrong place/not grepping hard enough .-. <lechner>PurpleSym / i am in way over my head, so any hints are welcome <lechner>well, thank you, sir, for your compassion and assistance! <winter>hmm okay looks like i just assumed there'd be a GCC wrapper <winter>i can't find any default CFLAGS, though <winter>so i think my question still stands <winter>unmatched-paren: yeah, checks out. do you know if there's a reason for this/why things like stack protection or PIE haven't been applied like other distros do? <winter>seems like an odd choice, if intentional <cel7t>I'm trying to build a package that uses imake, any suggestions on how to go about it? <cel7t>if I'm understanding correctly the normal C program compilation method won't work <cel7t>since some libraries have been hard-linked in the imake file <tux_life>Hi! What to do if you want to replicate a configuration, but what was declared in /etc/config.scm was not present in the Guix System installer iso? If I do a "guix pull" from the live environment, it runs out of space... <jpoiret>if you start the cow-store service you shouldn't run out of space (methinks) <tux_life>jpoiret I run "herd start cow-store /mnt" before "guix pull", but still... <jpoiret>my bad then, I thought it would've been enough <tux_life>I imagine that the simplest solution is to do a minimal installation via the installer and replicate the configuration later... <vivien>I’m back at trying to figure out how to build a flatpak with guix. The steps are conceptually very simple: build the package, obtain the closure of it, filter the static data that needs to be exported (icons, .desktop launchers, settings schema…), create a flatpak metadata file, stuff everything in the ostree repository, fix the branches and enjoy. Now, I expect a few pain points: how to convince flatpak that I do not require <vivien>any SDK or platform? How does ostree store symlinks? And more stuff that I’m not anticipating yet… <vivien>If you want to know just as much as me, install flatpak and libostree, install a flatpak to your home library, and explore the ostree repository at ~/.local/share/flatpak/repo. You can do: ostree refs --repo=$HOME/.local/share/flatpak/repo, ostree export --repo=… (ref) > somewhere/commit.tar, and then explore that tar. <vivien>In principle, ostree is a very interesting alternative to guix pack (for us), because it deduplicates identical binaries across packages and across remotes. <vivien>Since we claim to be bit-reproducible, that means updating an ostree should be very lightweight (remind me again how we are supposed to update guix packs?) <vivien>(that was a rhetorical question, the answer is we don’t) <awb99>I have guix 1.4 with wayland. audio is not working. "hwinfo --sound" shows me the device and it has a driver. but "pacmd list-cards" does not show anything <awb99>any ideas what could be the problem? <awb99> [pulseaudio] cli-command.c: stat('/gnu/store/l9886f9hhph8szl5pkkwri62rxn54kan-pulseaudio-15.0/etc/pulse/default.pa.d'): No such file or directory <awb99>Mar 8 16:44:09 localhost pulseaudio[5063]: [pulseaudio] bluez5-util.c: GetManagedObjects() failed: org.freedesktop.DBus.Error.ServiceUnknown: The name org.bluez was not provided by any .service files <awb99>this are error logs in /var/log/messages