<suitsmeveryfine>but I opened a new terminal and ran the command again and then I got only the second error <lfam>If the command didn't start with "sudo" then it was not run with root privileges <alezost>lfam: I just run dmd in my .bash_profile <lfam>I'm not totally sure but I believe that reconfigure requires root privileges. Try it with sudo <lfam>alezost: Okay, thanks. I wasn't sure if there was some magic or if it was as simple as that. <lfam>Do you redirect the output anywhere in particular? Just wondering if there is a "canonical" way <alezost>I redirect it to $HOME/.config/shepherd/shepherd.log (previously to $HOME/.dmd.d/dmd.log) <lfam>Alright, so no magic :) Cool <alezost>lfam: I start it like this: [[ -z $(pgrep -U $(id --user) shepherd) ]] shepherd &>> $HOME/.config/shepherd/shepherd.log <alezost>which means if it's already running, then do not try to run another instance <lfam>Do you think that belongs in the GuixSD equivalent of /etc/skel? <alezost>I don't think so, it's up to a user to decide if (s)he wants to use shepherd for user services or not <lfam>Are there any other service supervisors packaged yet? <lfam>I see pies (never heard of it, cool) and some skarnet stuff <suitsmeveryfine>lfam: it seems that sudo was required, yes. For some reason I get a linux-downgrade though (4.3.3) <lfam>suitsmeveryfine: It's possible that you haven't done `guix pull` since reinstalling? <lfam>Weird. What versions are reported by `guix package --show=linux-libre`? <lfam>Oh, did you `guix pull` as root? Each user has their own "view" of Guix. <alezost>lfam: I don't know about other service supervisors (and I've also just learnt about pies) <suitsmeveryfine>I did get an old kernel because I hadn't run `guix pull` as root so I ran `sudo guix pull` and reconfigured again and now I'm back with 4.4 <suitsmeveryfine>Yes, it's really nice. The next step is to try to enable suspend/resume <lfam>And hibernate! That's my favorite <suitsmeveryfine>Wow, it almost worked automatically. I just suspended and resumed. The only issue is that the mouse stopped working <suitsmeveryfine>It almost never works the first time. Well, I guess that I need to reboot now to get the mouse back <lfam>sneek: later tell suitsmeveryfine: I have to leave. Good luck! ***nckx|offline is now known as nckx
<Jookia>Is there a way to always build with debug symbols? ***francis7 is now known as incognito
***incognito is now known as francis7
***nckx is now known as nckx|offline
<civodul>ACTION .oO( It would be nice if M-x guix-challenge would spawn 'diff -r' on the differing copies... ) <Jookia>Is there a way to get an environment of a failed build, like 'guix environment coreutils'? <Jookia>It seems running 'guix environment coreutils' is kinda useless for debugging the 'coreutils' build failure as it just tries to build 'coreutils' again and fails <taylan>Jookia: 'guix environment coreutils' should not try to build coreutils, from what I know <Jookia>Maybe this is because I don't have coreutils built already and it depends on itself <Jookia>Passing --bootstrap doesn't help either <Jookia>civodul: I'll try with that branch <Jookia>civodul: That branch doesn't fix it <Jookia>civodul: Oh woops, I may not have switched to it? Git please <Jookia>civodul: Yeah, wip-environmental-profiles branch doesn't seem to fix it, oh well <Jookia>And running outside of guix environment in a post-build failure gives dozens of errors unrelated to the one I'm fixing including double frees :( <Jookia>Looks like I'm dealing with a bug in coreutils on btrfs :( <Jookia>And it's a bug on my system too, NixOS just disabled the test :| <Jookia>Actually, it's not a bug on my system! I wonder why it's a bug on Trisquel with the new coreutils <NiAsterisk>If I want to change/add some elements on the website (I am working on a collection of ideas which includes this), would I commit the patch to the devel list? <NiAsterisk>is there a way to build the website locally, so I can verify I do not create bad patches? <mark_weaver>NiAsterisk: yes, cd into the 'website' directory, run "guile -L ." from there, and then do (export-web-site "/path/to/dir") where /path/to/dir is a directory where the HTML will be rendered. ***kelsoo1 is now known as kelsoo
<NiAsterisk>hm. the build-system smart enough to figure out if it needs lua-5.1 or lua-5.2 ? <NiAsterisk>hm, recent awesome-wm requires lua-lgi. is somebody working on this or did fail before? there's luarocks but I read somewhere back in the logs from last year that lua is supposedly hard to package <taylan>I'll make a package manager that rolls virtual dice to decide what actual software to install on your system <mark_weaver>and here I thought packaging systems couldn't get much worse... <davexunit>the bright side is it's just an experiment. no one actually uses this. <civodul>lots of challenges for "traditional distros" <civodul>davexunit: BTW, could you add the slides of your talk to guix-maintenance.git and on the web site? ***nckx|offline is now known as nckx
<myglc2>I'm doing a fresh guixSD install on a dedicated SSD with only 1 partition, can anyone suggest the most copacetic way setup/partition the drive? <fps>i need to google that :) <myglc2>Sorry, the most likely to work. I had a problem with GRUB using the existing partitions so I thought I would just start fresh. <fps>myglc2: you could use fdisk and mkfs from the installation image <fps>hmm, what kind of problem? <davexunit>myglc2: I used GNU parted to make an ext4 partition that I installed GuixSD onto. <myglc2>My disk is gpt with bios_grub flags, detritus from a NixOS install. <davexunit>my disk is DOS, not GPT, so I'm not sure if anything special needs to be done there. <myglc2>Yeah, I was hoping to use GPT, but for NixOS I found I need the "bios_grub" flab <myglc2>OK, so I'll just use DOS for now, Thanks. <fps>DOS partition table format? <davexunit>I have never used GPT. I don't know anything about it. <calher>How do I install on Libreboot with FDE? <df__>SLiM is telling me "Failed to execute login command" and there isn't anything useful in slim.log - is there somewhere else I should be looking? <lfam>Is anyone working on the nginx security update? If not I can test the build and push <lfam>Anything keeping us on nginx 1.8? There is a 1.9 series <lfam>I'll do the 1.8 update and then we can think about 1.9 <lfam>Ah, that's why I asked ;) <davexunit>they don't call 1.9 unstable, actually, they call it "mainline" <lfam>Sounds like the same thing ;) <lfam>unstable and mainline, that is <mark_weaver>bah, another "high severity" flaw in openssl. we should try switching to libressl soon. <lfam>mark_weaver: It's supposed to have compatible API, right? So what would we need to do in order to switch? Just change all openssl referrers to refer to libressl instead? <mark_weaver>lfam: yes, but we'd want to do it on an experimental branch, and not now because we don't want to delay the security update in case of unexpected problems. <mark_weaver>or maybe it's something for the next core-updates cycle <myglc2>I just did a fresh guixSD install on a headless sever, how do I enable sshd? <mark_weaver>actually, I haven't researched the issue of API compatibility myself, but iirc that's what Andreas said. <lfam>mark_weaver: It seems that it has to be, otherwise nobody would use it. <mark_weaver>myglc2: see 'lsh-service' in the "Networking Services" section of the manual. you need to add lsh-service to the list of services in your OS config and run "guix system reconfigure" <myglc2>Thanks Mark. I used the bare-bones config as a starting point, and it has lsh-service already in the "baseline". So I think I aready did that, but I can't ssh into the machine. <bavier>myglc2: I'm not a user, but I recall hearing that lsh requires some sort of setup before it can actually be used. <calher>oh crap, can i torify servers on guixsd? <calher>Oh wait, nevermind. My server will run Trisquel and my laptop will run GuixSD. <myglc2>Thanks bavier, that is probably what I am looking for. <mark_weaver>lfam: please let me know when you've pushed the nginx security fix. I want to deploy it to hydra asap. <lfam>mark_weaver: It built. I'm trying to test the resulting executable but I can't get it to respect my configuration of where to put the error_log. It wants to put it in the store which won't work of course. Testing nginx.conf: http://paste.lisp.org/+6K2L <lfam>I can push if you think this is not important <mark_weaver>lfam: did 1.8.0 work for you with the same configuration? <mark_weaver>i.e. do you have reason to believe this is a new problem, or did you never have this working? <mark_weaver>unless you know it worked before for you with the same configuration that you have now, I would just push it. <zacts>I'm running my first 100% free software cell phone with replicant <zacts>I'm running guix on-top-of debian for now, but probably this weekend I'll be running guixsd fully <zacts>the last time I tried I forgot to add the initial disk sector for grub2 + gpt <zacts>and life got busy with my classes <zacts>mark_weaver: bavier: thanks! :-) <zacts>but it's only a USA startup for now, but soon Canada, Mexico, and Europe hopefully <zacts>they work really well with replicant <bavier>I was just reading about ting yesterday <zacts>I prefer Ting over T-Mobile, Cricket, and Google Phone for sure <zacts>bavier: I discovered them initially from the Linux Action Show podcast <zacts>everything works except for bluetooth and wifi, but 3G celluar network works fast <mark_weaver>zacts: what is an important advantage of Ting over T-Mobile? <zacts>mark_weaver: I think customer service, and lack of need for non-free blobs to connect to their cellular service <zacts>I think t-mobile wants monitoring apps, and things to use their service <zacts>(I need to double check on this, but I needed several T-Mobile apps on a previous phone I think) <lfam>zacts: Does Google bake those apps into Android? I'm using a stock nexus device on T-Mobile <zacts>lfam: some of them they do. T-Mobile TV I could not uninstall without jailbreaking the phone <zacts>(thus my inductive logical conclusion, which may prove to be false) <lfam>I think that must only apply if you buy the phone from T-Mobile. I don't have anything like that except for the visual voicemail app which I installed myself <mark_weaver>Is it not possible to use a phone running Replicant and no additional non-free blobs with T-Mobile? <zacts>but I like ting because of BYOD, but also the web app for monitoring my usage, and it uses only what you use each month, and the base fee is only $6 per month for the phone <zacts>so if I use no texts one month, I don't pay for them <lfam>Hm, that's interesting. I've been looking for something cheaper than T-Mobile. <zacts>mark_weaver: I'm not certain, it may work with T-Mobile. I don't mean to spread misinformation at all. (disclaimer: Inductive Logic was the way I based my conclusions) <mark_weaver>If Ting is truly more free software friendly than T-Mobile, I'd like to know, but I'm dubious. <lfam>I don't think any of the cellular carriers are not 100% user hostile. But if they let you BYOD then that is good. <zacts>I can get an overall lower rate, but the data usage is kind of pricey <zacts>but I also like supporting a company that aims to provide fiber internet, that isn't google <zacts>they seem more co-op like (disclaimer: again, this is inductive logical) <zacts>and I'm trying to get it brought to my city <lfam>It looks like ting would be about the same price for my max usage, but it could be good on my slow months <davexunit>I use t-mobile and I brought my own phone when I signed up. <mark_weaver>I've always used my own (dumb) phones with T-Mobile. <lfam>I always think of DJB's "I am the man in the middle" presentation when these issues come up <lfam>IMO there's literally nothing cellular providers won't do to make money off you <zacts>mark_weaver: but anyway, I'll try to gather the _facts_ on ting, and I'll let you know <mark_weaver>T-Mobile seems to have much better policies than the other major providers in the US, but of course that's not saying much. <mark_weaver>by major providers, I mean the ones that actually control their own towers instead of being forced to rent other provider's infrastructure. <lfam>I was thinking of switching to metropcs since they are owned by tmobile but seem to provide the same service for less money <myglc2>guixSD install failed, but succeeded with the --fallback option. The same thing occured with 'guix package -i emacs' Is this normal? <lfam>myglc2: Not really. What messages did it fail with? <myglc2>unfortunately I was running in tty console, so can't say. now that I have emacs I can run in a buffer and see what is going on. <myglc2>the good news is that 'guix package --fallback -i emacs' succeeded after much huffing and puffing (~15 min) so now I have emacs, without which I am an abject incompetent <lfam>It would be good to know because that usually shouldn't be necessary. And I believe that in some cases, it being required indicates that some build is nonreproducible (failed on hydra.gnu.org but succeeds locally) <myglc2>OK I will have more specific info next time. <mark_weaver>I wouldn't be surprised if substitutes from hydra are timing out now. hydra is not happy right now. <mark_weaver>hydra is (over)due to be replaced, and hopefully will be within a couple of months. <nckx>ACTION was just installing his very first Guix and was a bit disappointed... Thanks for the info. <CompanionCube>ACTION makes a mental note to put off trying out guix until hydra is faster/more stable <lfam>Give it a little while. The bandwidth comes and goes throughout the day. <mark_weaver>at the moment it's doing a much-needed garbage collection, and also at the same time I'm deploying a security update to it. this is especially bad right now. <lfam>You don't need to wait the few months until we get a new server <CompanionCube>I can only imagine the CPU usage of a build farm that's constantly building and rebuilding things <nckx>TIL: there's no binary S3-alike cache for Guix. <myglc2>Oh I wouldn't wait... this is not to be missed. I spent a week trying to grock the Nix language and completely shorted out. this looks so much more understandable, at lease IMHO <lfam>CompanionCube: This isn't even the build machines, just the front-end <mark_weaver>the front end is a woefully underprovisioned VM, with not nearly enough RAM or disk or disk bandwidth. <mark_weaver>we have grown, our needs have grown, and there was no room to grow on this particular VM. <mark_weaver>so we'll be moving to a vastly more capable bare metal machine. <nckx>myglc2: how HMOs can differ... I can't grok this at all. Yet. Still. Mainly curious about the Scheme-replaces-bash aspect. <CompanionCube>I'm looking more into Guix than Nix because I can see the advantages of using a turing-complete actual programming language for configuration and description <myglc2>I live in emacs and have always had trouble with {};'s so it might just be the soothing effect of these ()'s. <nckx>CompanionCube: I'm pretty sure that Nix is theoretically Turing-complete. Which isn't the same as usably. <myglc2>I am not a language purist, but I appreciate that guix is using guile and this seems like a more leveragable thing to learn more about. <nckx>CompanionCube: no disagreement. I'm just curious about how maintainable/disciplined a package repo in Guile is. And how laziness is achieved. Which is why I'm going to try it out in *checks installation progress*... oh, 2018 or so. <CompanionCube>doesn't guix fallback to compiling all the things locally if substitutes cannot be found / timeout <davexunit>nckx: what type of laziness are you referring to? <lfam>CompanionCube: If there is no substitute available, then Guix will compile. But if the substitution fails for some reason, then you must specify "--fallback". That is my understanding. <myglc2>Certainly did a bunch of compiling just now to get emacs installed. but hey, I like seeing that work and compute cycles are basically free. <lfam>myglc2: Yes, the substituter most likely failed due to network issues. Although it would be good to know why. <myglc2>lfam: so the next time I see this I will report some "possibly useful" details <a_e>myglc2: This is quite certainly just a timeout because hydra is too slow right now. Nothing we could do at the moment. <mark_weaver>bad timing that hydra was already very busy with GC when I found out about a security update to deploy to it. <myglc2>a_e: please understand I am not complaining. emacs is installed and running and I actually don't care about the delay as long as it doesn't mean something is wrong on my box, heh. <davexunit>I'm not sure what laziness you feel needs to be achieved. <a_e>myglc2: Sure. I just meant that you need not spend your time reporting an issue. <nckx>davexunit: I feel nothing. <lfam>myglc2: The transparent substitution of binaries (when they are available) is a very nice feature in my opinion :) <davexunit>nckx: the Nix language happens to use lazy evaluation, but that doesn't mean that it's the only valid evaluation model for a functional package manager. <mark_weaver>Scheme is eager by default, but it also supports laziness where explicitly requested. <myglc2>lfam: oh yeah, and I saw it working great yesterday. <nckx>ACTION 's current knowledge of Scheme: oh look parentheses <CompanionCube>ACTION originally disliked the lisp family for parens everywhere but no longer cares <lfam>Dealing with the parentheses is trivial with tools like paredit <nckx>Editing my first config.scm in nano was fun. <jonsger>I'll hope my donation will speed up hydra, 25KiB/s :( <fps>df__: i looked for failure details or slim, too, at one point in time <fps>i just went with creating .xsession and symlinking it to all kinds of places like .xsessionrc, .xinit and .xinitrc and start my stuff manually there <fps>note that .xsession is marked executable <mark_weaver>okay, I finished deploying the security update to nginx. <myglc2>This is a total turn-on: emacs comes with 'GNU Guix' INFO <CompanionCube>Guix comes with a emacs interface for managing your packages <myglc2>Oh yeah, when I saw that in the video, I really didn't have any choice. <myglc2>running guisSD, where is the best place to put /etc/hosts-type info?