<rekado>Wechat is one of the worst apps out there, and unfortunately one of the most influential as well. <jlicht>bad jokes aside, it frightens me how people can be faced with how their choices contribute to systems such as these, and still shrug it off as 'not my problem' :/ <helaoban>I lived in China for a long lime, an interesting observation was that many kids these days completely bypass traditional means of online communication and move straight to wechat. They will have never used email, and SMS has been relegated to a kind of spam-zone that makes it unusable. Contacts are rarely saved in contact books anymore since there's almost never a reason to drop to the OS to do <helaoban>anything, e.g. very few calls are made directly through the phone over the cell networks, and are made mostly through Wechat's call feature. <helaoban>ryanprior: so after skimming the docs for time-machine and inferiors, it looks like the right mechanism to use packages from multiple different guix revisions is inferiors, is that right? Say for example that I wanted to use python 3.7.4 and elixir 1.10.3, both of which are no longer in the current revision. <ryanprior>Absolutely, that is the intended use of inferiors and that is the easiest way to accomplish what you want. <ryanprior>What you said about communications in China resonates too, I've never lived there but know a number of Chinese students in the US who only communicate with friends and family in China through WeChat. <ryanprior>Calling it a chat app really undersells it from what I'm told, it's an app store with tons of games, it's one of the most common ways to pay for things, even food and clothes etc <ryanprior>It's like the apotheosis of proprietary software surveillance capitalism ***j is now known as jess
<helaoban>ryanprior: cool, I'll play around and see if I can get an environment up. <helaoban>Yeah, payment is the other big use. The unfortunate reality is that it's very good at what it does. Ethics aside, I admire some of the purely techincal decisions they've made, for example using QR codes and cameras for payments instead of RFID. <helaoban>QR codes were already super popular 10 years ago in China as a marketing tool (think giant QR codes on the sides of buildings) and they just hijacked that to bootstrap a payment system. <helaoban>So as long as you have a phone capable of displaying a QR code, a cashier could scan it at a POS machine and the transaction is effected immediately. And if you were buying from a food stand that couldn't afford a POS terminal, then you would just scan the merchant's QR code which he has pasted everywhere on his cart and you're good to go. <helaoban>I watched China go from cash to cashless in a span of ~4months, it was crazy. <nckx>Does anyone know/remember why hplip is built with --disable-network-build? scan/sane/hpaio.c now hard-depends on it, which might be a mistake, but I wonder if there's anything inherently problematic about just enabling it. *nckx 'll ask again when #guix is awake. <gnutec>jlicht: I'm talking about the app in android. It open a web browser to inscription. ***jess is now known as jess-o-lantern
<bdju>is there a process for just flagging something out of date when I notice it? the freerdp package is several versions behind <roptat>the best process is to send a patch ;) <roptat>not sure if a bug report is a good idea or will simply be annoying ***catonano_ is now known as catonano
<joshuaBPMan>Hey Guix, I'm trying to get my guix server set up to serve my email. Does anyone here use spamassassin? I don't see it packaged in guix. <roptat>joshuaBPMan, I don't think we have it ./ <roptat>I have no spam filter on my email server <joshuaBPMan>roptat really? I guess I am just imaging that you get a lot of spam... <roptat>but usually it's easy to filter, it's in cyrillic or Chinese for some reason... <helaoban>is there an easy way to search guix for the last commit with a given package version, without having to clone the repo and searching the git tree myself? <roptat>helaoban, maybe ci.guix.gnu.org has that info <ryanprior>Oh cool do I just send an email that says "close"? <Brendan[m]2>and include in the message a short message saying its a duplicate <ryanprior>Okay that's taken care of, thank you Brendan! <luhux>Hello everyone, how does Guix boot using PXE? My machine has a problem with USB, it will report an error during the copy to /mnt stage and then init will be killed. I am looking for another installation method <efraim>xelxebar: good news is the pine64 came back up, bad news is there seem to be some bugs with creating the install image <sneek>Welcome back efraim, you have 1 message! <sneek>efraim, xelxebar says: Hey, just a quick ping. Any news on what happened with your pine64 board? <efraim>there was some bug with the v86d service and non-Intel machines logic, I forget what the second one was <g_bor[m]1>With all these nice additions on netbooting, I was wondering how lightweight is to add the ssh public keys to different kind of system targets <efraim>xelxebar: there's also bug#43591 which means there might be problems using aarch64 to build an armhf image anyway <g_bor[m]1>I have a provisioning system in mind where a lightweight guix system image is generated, basically differing only in the provisioner user ssh public key for the nodes, that pxe boot it, and then the final config is injected with guix deploy. Anyone doing something like that? ***lewo` is now known as lewo
<Brendan[m]2>When I run guix environment --ad-hoc guile3.0-chickadee, it does not prepend to GUILE_LOAD_PATH or GUILE_LOAD_COMPILED_PATH at all, so the module cannot be found <Brendan[m]2>looks like it works if i add guile its self. interesting <rndd>little newbe question. how read-only filesystem is done in guix? where i can read about it? <nckx>rndd: (guix)File Systems <nckx>There's a difference between flags and options, although the command-line ‘mount’ combines them in one list. <nckx>Brendan[m]2: That's how search paths work: the consuming packages declares them and causes them to be set in the profile, not each provider. <nckx>liltechdude: I don't know much about Racket but note that the sfml package provides libsfml-audio.so, not libcsfml-audio.so. <rndd>nckx: thanks, will read it ***robmyers_ is now known as robmyers
<efraim>note the looks-like-guix user handle, followed by going afk immediately, follwed by logging out <nckx>It's true that our recent spammer was not gud with computer, but even they would use a real TLD, and probably haven't moved to Siberia, much as I would appreciate that. <nckx>raghavgururajan: I was hoping that you could explain the joke to the rest of us, but alas, I guess. <nckx>Hey now, please don't doxx people even so very mildly. <nckx>IPs are PII & you'd get in trouble in places. <OriansJ`>raghavgururajan: IPs are not meaningful idenifiers for trolls or spammers as they can be bought as a service. <raghavgururajan>nckx: Hello to you too. What joke? I thought guix-vits 's nick was hijacked again. <nckx>OriansJ`: And vits is neither, making it doubly irrelevant. <nckx>Why did you think that? I don't follow either. <nckx>Where is the spam? Where is the hijacking IP? <nckx>liltechdude: #:make-flags (list (assoc-ref %build-inputs "the package")) ; and add the package to an *inputs field. <raghavgururajan>nckx: It's just that, there was one time the spammer posted under the nick guix-vits, when *real* guix-vits logged out. <nckx>liltechdude: There isn't currently a direct equivalent to that so it depends on the context. You'd use a slighly different notation in a phase. *raghavgururajan goes back to patrolling the premises <Brendan[m]2>liltechdude that is done by using (assoc-ref inputs "smfl") ***OriansJ` is now known as OriansJ
<Brendan[m]2>you can see many other examples for how to make inputs defined <nckx>Brendan[m]2: ‘inputs’ needs to be bound for that to work. <OriansJ>raghavgururajan: well every users' name can be claimed by anyone; that is why registration and msg nickserv ghost USERNAME PASSWORD are a thing <nckx>raghavgururajan: I just don't see the spam in that message? I can assure you that it wasn't. Implying that people are spammers merely because they're... idiosyncratic triggers me beyond reason, it is true. In any case: thank you for your service! o7 <OriansJ>raghavgururajan: so never trust a name because I could be anyone from anywhere trying to do anything... <raghavgururajan>nckx: I agree. The the message plus efraim's comments triggered me into thinking it was spam. Okay, it is not. :-) <nckx>It is a crypto puzzle for greater minds than ours. <nckx>Stop pasting bootstrap porn. <Brendan[m]2>i suppose they can still be used if one already has a hash for them <OriansJ>nckx: but I wrote the world's first C compiler ever written in ARM assembly <raghavgururajan>> OriansJ: nckx: but I wrote the world's first C compiler ever written in ARM assembly <nckx>Brendan[m]2: The idea is that they are simple enough to be audited independently by regular humans, and don't need to be blindly trusted. <OriansJ>Brendan[m]2: do you know of any secure has smaller than the 240 byte binary? <nckx>Brendan[m]2: Yes, that's inevitable. <OriansJ>raghavgururajan: yes I am Dad with good Son <nckx>‘do you know of any secure has smaller than the 240 byte binary’ - this is your brain on ARM assembly, kids. Not even once. <raghavgururajan>OriansJ: Glad! I was making a joke that I claimed someone my dad who wrote first c compiler .... <OriansJ>nckx: amen to that; ARM assembly requires an insane number of stupid work arounds *raghavgururajan goes to play with racoons <OriansJ>nckx: it is also 6:30am and I am just starting to coffee myself awake after working until after midnight last night. <nckx>OriansJ: My sympathies. I hope it was somewhat voluntary. <Brendan[m]2>if i run nheko with QT_QPA_PLATFORM=wayland-egl i get an error saying that backend doesn't exist <nckx>Brendan[m]2: No, that does sound like a bug in the wrapper, since even if QT_PLUGIN_PATH is set in the surrounding environment they'd still overwrite it with ‘=’, right? <nckx>‘No’ was not in response to raghavgururajan, maybe that's the solution. <OriansJ>nckx: my son Thomas is almost 7months old; so not entirely <nckx>bdju: Are you updating freerdp? <nckx>OriansJ: Gah. They grow up so slowly. *nckx sure to grow up any day now. *raghavgururajan does *pikaboo* for nckx <mroh>my oldest son is 24 now, kids grow up very fast... <OriansJ>mroh: I believe the phrase is that the days are long but the months(/years) are short <OriansJ>because it still feels like yesterday he was born and boom; we are at 7months already <rekado>my baby just turned 1; my level of sleep deprivation and exhaustion is entirely consistent with that <OriansJ>rekado: I don't know about you but the first night my son slept 6 hours straight was the best night of sleep in my life. <rekado>our portable human woke up once every hour last night (she’s sick and so am I); when she sleeps quietly for too long, though, I feel the irresistable urge to wake up and check on her. <rekado>I do like that she will sleep in the afternoon for about 1.5 hours. That’s when I can either decide to do a few things or catch up on sleep myself. <OriansJ>rekado: I do hope she feels better soon. Knowing one's child is sick just feels so much worse now that I am a parent. <rekado>thanks! It’s bad that they don’t know much at all, so they don’t have any good coping strategies yet. <OriansJ>rekado: it is handy that they cue off your reactions; basically act like everything is fine and they just calm down. <rekado>sometimes ignorance is not bliss. But a poor memory sure is! <rekado>pity it doesn’t work quite as well for falling asleep itself <OriansJ>although the 6 month sleep regression threw me for a loop. Who knew forcing daytime naps resulted in faster return to full nights of sleep. <OriansJ>and routines and schedules help reduce the stress which prevents proper sleep. <OriansJ>Now I have a 10 hour track of rain sounds that cue night sleep and a 4 hour long techno mix that seems to help my son sleep in the morning after he wakes up. <OriansJ>still attempting to figure out the magic music combination to getting him to sleep in the afternoon though. <OriansJ>but the music styling of a Tribe called Red; does seem to have a useful calming reaction from him <rekado>for us the cue to sleep at night is a playlist of Billy Cobham’s Spectrum, Mahavishnu Orchestra’s Lotus on Irish Streams, and (if necessary) Autumn Leaves. <rekado>after the afternoon nap it’s all Vulfpeck and Bach concertos for rocking out <rekado>to veer back to Guix: Bach concertos are great hacking music, too! <OriansJ>rekado: I find them too distracting and tend to play trance as it helps empty out the world around me. <rekado>(or so I hear… I can only ever do one thing: listen to music or hack) <nckx>Bach is too interesting. <nckx>bdju: ...so I went ahead and pushed a freerdp update because of CVE-2020-15103, although I forgot to mention that in the commit message. Apologies if you were working on your own. <mfg>how do computed-files actually get written to disk? *raghavgururajan + Zopiclone = zzZ <ryanprior>I'm checking the output of one of my packages to make sure I have the right inputs declared, and I notice it's linking to my system libpthread and libc. How can that be? <mfg>OriansJ: i mean objects returned by computed-file; i'm trying to understand how i.e. $GUIX_PROFILE/etc/profile gets generated <OriansJ>mfg: by a forest of symlinks; essentially /gnu/store is just a big bag of binaries. The profiles are just collections of symlinks to the binaries needed and the current profile just points to one of the collections of symlinks <cbaines>ryanprior, I believe there's some special stuff somewhere in the toolchain Guix uses to have dynamic linking target specific store items <cbaines>I'm guessing this is a Go binary, so that might behave differently... I'm unsure <bavier[m]1>ryanprior: it's a dynamic library, so `ldd` will "link" to whichever library it can find. Here that probably means there's no rpath embedded for libc and libpthread when the library is built <cbaines>mfg, a profile is an item in the store, like /gnu/store/bs85y2q3zdzmxmx7bwykjr6zv12znm2x-profile There will be a derivation to create it, and something describing that derivation in Guix (it could be a computed file, I haven't checked) <cbaines>mfg, specifically about computed-file, that's a high level thing from which a derivation can be generated <cbaines>mfg, so "how do computed-files actually get written to disk?" <cbaines>I would say, a derivation is generated from the computed-file, that derivation has a build script, and when that's run, it does the writing of any files <mfg>cbaines: that makes sense, thank you :) Now i now what i have to read :D <ryanprior>cbaines: it is indeed a go binary. bavier should I patch the binary somehow so that it has an rpath embedded, to make the package more correct? Or is it working as intended? <jlicht>ryanprior: I had a quick read over your hugo-deps earlier today; <jlicht>I don't think I'll have a lot of time in the upcoming week for a more in-depth review. My apologies ryanprior, you'll have to poke 'n prod someone else for that if you want it to happen sooner rather than later <ryanprior>jlicht: no problem, I opened an umbrella issue yesterday in the mailing list and sent my first patch series of Hugo deps. <ryanprior>I started with just 3, gonna trickle them out like that and apply what I learn from feedback to others I haven't sent out yet. <cbaines>ryanprior, I'm unsure. it might just be an unaddressed issue with how golang things are built in Guix <cbaines>ryanprior, do you think it'll block you trying to package what you're working on? <bavier[m]1>ryanprior: I'm not super familiar with go tooling, but based on some other packages, it appears you can set and `LDFLAGS` environment variable? <ryanprior>cbaines: it's not blocking me, just trying to submit correct patches upstream :) ***ping is now known as niko
<luis-felipe>Man I'm frustrated with Guix. Very often, when I install something new, I can't try the thing out for some error or omission. <luis-felipe>Sorry to be whining here. I appreciate all the work you all do, but it's really frustrating... <leoprikler>looking at that error it seems as if $prefix is not correctly expanded in the forming of this path <leoprikler>yep, it's in the line #:configure-flags '("-DDATADIR=share/astromenace")") <leoprikler>you probably want (list (string-append "-DDATADIR=" (assoc-ref %build-outputs "out") "share/astromenace")) <bdju>nckx: was not updating it myself, no. thanks for the update <Inquiz>New here. An I correct that Guix is a project to write GNU Utils in Guile? <luis-felipe>leoprikler: Thanks for looking. I would patch it if I new how to do it. I've packaged some things before, but I don't have the background to provide high-quality, well tested packages. <leoprikler>luis-felipe: the manual goes into more depth, but it's basically downloading the source code, finding the location, editing, testing, sending a patch to guix-patches@gnu.org <mfg>are there some tunable options to make guix time-machine faster? my guess is the only way to make it faster would be to throw more cpu ressources at it, because most of the time it's busy compiling things? <zimoun>nckx: about “shell completion installation” files in guix-install.sh and bother to the user, it appears to me just the polite way to ask and not install stuff behind (even about profiles). So IMHO, 2 options: update the manual to clearly state what guix-install.sh is doing (for example in the Warning note) or add only one question directly in guix-install.sh for profiles and shell completion. WDYT? <sneek>Welcome back zimoun, you have 2 messages! <sneek>zimoun, rekado_ says: issues.guix.gnu.org has *access* to all GNU Debbugs bugs. It doesn’t include them in the search, nor does it index them. But it can show them all. I think that’s a feature, not a bug. <sneek>zimoun, rekado_ says: issues.guix.gnu.org could refuse to show them, but I think that would be a needless limitation. <lafrenierejm>I'm getting the following error on all `guix import` invocations: <lafrenierejm>In procedure struct-vtable: Wrong type argument in position 1 (expecting struct): #f <ryanprior>Inquiz: the Guix project is written in Guile, and it implements many features similar to what are available in other GNU utilities, so that's not necessarily wrong <ryanprior>But the purpose of Guix is not to replace other GNU utilities with Guile, but rather to provide a nice tool set for describing packages, services, and whole systems using Guile <rndd>hi everyone! what is propogated inputs? <ryanprior>And then doing useful things with them like installing, managing, deploying, testing, publishing, etc <ryanprior>rndd: if you install some package P, all of P's propagated inputs are also put into your profile. <ryanprior>P's inputs are put into your store, and P might make use of them via direct links, but they would not be in your profile <ryanprior>That's the difference between inputs and propagated-inputs. <leoprikler>IOW propagated inputs are "propagated" into the profile a package is installed in ;) <civodul>lafrenierejm: hi! could you give an example command to reproduce it? <efraim>guix import pypi black-macchiato <efraim>that said it doesn't fail for me <mfg>is it possible to boot guix system over pxe, mount the store via nfs and have the rest of the system in a ramdisk? so basically run guix diskless? <joshuaBPMan>man, it is not really easy to debug why a guix service fails to start. It's also not really easy to discover where the latest configuration file is. <sneek>Welcome back joshuaBPMan, you have 1 message! <sneek>joshuaBPMan, nckx says: We don't have SA (is that still widely used?) but we do have Bogofilter. I also have an unfinished rspamd patch somewhere. It basically works but turned out too heavyweight for my usage: redis? Not when a simple Sieve rule matching the Greek alphabet + "funds" catches 90% of my spam... <joshuaBPMan>sneek: I'm surprized than a simple sieve rule catches most of your spam. <joshuaBPMan>I've manually tracked down my opensmptd conf file via: <nckx>zimoun: Neither option sounds good to me, but I'd just be repeating my mail. Why would installing a command's shell completion alongside it require a warning? No other package does this. <nckx>(joshuaBPMan: sneek is a bot.) <nckx>It's probably closer to 95%. <zimoun`>nckx: Ok. I disagree, but fine I will send an updated patch removing the question and just installing. <nckx>zimoun`: You don't have to agree or adopt my suggestion if you don't want to. <nckx>But you gave me 2 options that I don't agree with: I couldn't well choose 🙂 <nckx>It wasn't a rhetorical question above. <plattfot>Hi, everyone! I'm having some issues building guix from git and I'm wondering if anyone could help me figure out what I'm doing wrong. Thanks. <mfg>what's the problem plattfot? <zimoun`>nckx: to be honnest, I do not have a strong opinion on the topic. Since you are reviewing (providing feedback) and I want to move forward because it is “boring” (damned shell! ;-)), so let’s do how you are proposing; you are surely wiser. :-) We could change later if there is bug report and/or “complaint”. BTW, you are right about ’nscd’ and init system, I <zimoun`>am updating the patch too. Thank for your feedback. <plattfot>mfg: I've followed the steps in the manual, i.e. cloned guix, ran guix git authenticate then guix environment guix --pure; ./bootstrap; ./configure --localstatedir=/var; make check <mfg>plattfot: i guess you are following the manual? so this happens during the 'make check'? if you are on the newest revision, then i would open a bug ticket for it. <mfg>s/i would/you should/ XD <plattfot>Yeah, I'm up to date with origin master. So I'll open a bug ticket for it. Thanks <nckx>zimoun`: It's probably subjective (of course I disagree, but... that's what the subject always does :-) - I'd never suggest enabling or starting the nscd service unasked, for example. <nckx>plattfot: Are you using bash? Assuming you are, if you explicity run <nckx>export SHELL=`which bash` <nckx>and the test suite again, does it change anything? <nckx>Of course ./configure should set CONFIG_SHELL regardless but maybe it's confused. <plattfot>nckx: I'm running zsh so maybe that's why <mfg>plattfot: also: is this on guix system or on a foreign distro? <nckx>plattfot: Also, when you write that your ran ‘guix environment guix --pure; ./bootstrap; ...’ - can we assume those ‘;’ are shorthand for ‘and then I ran’? If you really typed ‘guix environment guix --pure; ./bootstrap‘, that won't work. <nckx>plattfot: Hm, that might be a factor. Still a bug of course, but do mention it. <efraim>hmm, librsvg-next needs rust-1.40 <mfg>does the configure script explicitly search for bash? <nckx>Does we assume ‘sh’ is ‘bash’ somewhere? The answer is probably ‘probably’ :-/ <efraim>I normally use: guix environment guix -- sh -c './bootstrap && ./configure --localstatedir=/var --sysconfdir=/etc && make' <nckx>config.status:SHELL=${CONFIG_SHELL-/run/current-system/profile/bin/sh} <nckx>plattfot: What does yours say? <nckx>The above is on a Guix System, yours will be totally different. <nckx>But it shouldn't end up being "". <efraim>On Debian I have: SHELL=${CONFIG_SHELL-/bin/bash} <nckx>plattfot: It's probably a good idea to try something like efraim suggests. It's how I've been bootstrapping Guix for years. <plattfot>nckx: mine says SHELL=${CONFIG_SHELL-/bin/sh} <nckx>Maybe throw in an env -i $(which guix) … for even moar purities. <plattfot>I'll try efraim's suggestion and see if that works better. <plattfot>what's the best way of nuking all the config stuff and start over? Just `make distclean`? <nckx>or make clean distclean just to make sure (I can never remember which one implies what and it's not obvious) *nckx just nukes all the git-untracked files. <efraim>i normally end up with 'make clean' or 'make clean-go' and maybe have to hunt down leftover .go files <efraim>cdcbb42aa398f697dfe86563b98bddc3cb3d525a <efraim>uh, bump icecat to 78.3.1. about HEAD~30 <lafrenierejm>efraim: Thanks. I'll install from that commit and see if it works. <nckx>efraim: Yeah, something like that (‘C-r git cl’ to be honest...) <nckx>Gets rid of those annoying *.go.?????? files too. <efraim>lafrenierejm: my only idea is maybe you're missing SSL variables, what do you have for SSL_CERT_DIR and SSL_CERT_FILE <lafrenierejm>efraim: Yeah, I also don't suspect SSL. The only relevant (AFAIK) environment variable I have is SSL_CERT_FILE=/etc/ssl/certs/ca-certificates.crt <plattfot>nckx, efraim: `make distclean` and then run `guix environment guix -- sh -c './bootstrap && ./configure --localstatedir=/var --sysconfdir=/etc && make'` seems to work. It passed tests/store-roots.scm. <nckx>plattfot: I think it's still worth filing a bug, with details about your system & what did & didn't work (with the error), if you're willing to do so. <plattfot>nckx: Will do. Just going to do some more experiments to see what works and what doesn't. <nckx>Oh lord, rust-is-executable is the left-pad of our times. <jonsger>is there a way to stop an evaluation of cuirass? 