<rekado>raingloom: yes, efraim answered. The certificate on bayfront.guixsd.org has expired. Should be fixed soon. <raingloom>rekado: thanks! I'll check the log in a few hours then <rekado>raingloom: here’s what efraim wrote: <efraim> raingloom seems to be out so someone tell them that IIRC ada currently can't be bootstrapped, it needs an ada compiler to build for ada, so we unfortunately currently don't have ada support in guix <raingloom>rekado: dang. guess I'll do my Ada-ing on Arch then. although that doesn't explain the error guix was giving. <raingloom>oh wait... maybe I should have used an absolute path... <vagrantc>hrm. looks like there's a GPL/Openssl incompatibility in guix's u-boot packages ... interestingly enough, i believe the simple workaround is to mark the effected packages with #:substitutable? #f ... since the issue only comes up with binary distribution <vagrantc>many of the u-boot targets don't actually link against openssl, or can be made not to (this is done in debian) <vagrantc>guess i should probably file a bug about this <vagrantc>there are possibly other affected packages <vagrantc>gah. this issue is so muddled and hard to search for any good information on it <mikadoZero>For a package like emacs-lispy what does the revision before the commit hash mean? Revision number 1 in this example. <leungbk>"go: disabling cache (/homeless-shelter/.cache/go-build) due to initialization failure: mkdir /homeless-shelter: permission denied" <leungbk>i've tried messing around with environment vars to make that go away, but i still get this error. <leungbk>has anyone else had to deal with this before? <vagrantc>is it trying to download some dependency not present in it's inputs or something? <leungbk>i don't think so, cloning the repo myself and running "go test ." works fine <leungbk>i've seen some other guix package defs that have homeless-shelter related errors, and i've tried to mimic my workaround to match those. <vagrantc>my hunch is the homedir issue is unrelated to the test failure, since the build completes successfully after the error <leungbk>i might just patch out that test; apparently debian had a similar issue with this package. <ngz>Hello. For wireshark package update, I think I need to augment RPATH with "/gnu/store/...-wireshark/lib". Is there a usual Guix, or CMake way to do that? <ngz>In fact, it seems to use $ORIGIN in RPATH, but I don't know how Guix supports it. <rekado_>ngz: powertabeditor sets some RPATH related variables. Maybe that’s similar? <rekado_>mikadoZero: git commit hashes don’t increase monotonically, so a more recent version may be sorted before a previous version. We use the revision to ensure the sort order. <rekado_>the ordering is important as packages will not be upgraded when the new version string is “smaller” <ngz>rekado_: Yes it could be the same symptom. <ngz>It doesn't lose the full runpath, though. Only the lib/ subdirectory is missing. <rekado_>there are many examples in gnu/packages where the lib directory is added to the RPATH. <rekado_>I haven’t seen any using the cmake-build-system, though. <ngz>If I try the same trick as powertabeditor, all tests fail. So I guess this is not the correct approach. <ennoausberlin>Hi. I gave it another try to install guixsd 0.16, but I am stuck again <ennoausberlin>I just can not boot the image provided as usb stick iso. My uefi sees the stick but instead of booting its right back to the uefi menu <ennoausberlin>if I try to boot this from an existing grub (from ubuntu) it boots the kernel but at the mounting root fs step it claims about no such file or directory <ennoausberlin>what drives me nuts is that the partition scheme looks completely broken. Always a hfs partition is recognized after I wrote the image via dd if=./guix...iso of=/dev/sda <kmicu>Hi ennoausberlin: some sanity checks: did you check with gpg whether downloaded image is not corrupted? Did you ‘sync’ before removing USB stick? <ennoausberlin>I can not explain this behaviour, but some already told me, that the .iso image is correct <ennoausberlin>I have to download it again, because I already unzipped it with xz and the original is gone <kmicu>(We can use ‘xz --decompress --keep …’ to avoid that.) <ennoausberlin>I didnt use gpg for a long time in lack of participants. So I am not sure if have to configure it beforehand <ennoausberlin>Ok . Its finished with gpg: keyserver receive failed: No data <ennoausberlin>gpg --keyserver pool.sks-keyservers.net --recv-keys 3CE464558A84FDC69DB40CFB090B11993D9AEBB5 does not work <kmicu>The very same command work here. So I assume some networking issues. <kmicu>Let me check sha256 for the image. <kmicu>sha256sum guixsd-install-0.16.0.x86_64-linux.iso.xz <kmicu>07cef601a1d24936d0a1ea11a9ae905550aba650baa2230b1da46018b7ff97f4 guixsd-install-0.16.0.x86_64-linux.iso.xz <ennoausberlin>Can you mount loop the ISO and tell me the partitions you see? <kmicu>Let me check if that stick works on a laptop with UEFI cuz I’m certain it works on legacy BIOS. <kmicu>BTW did you test any other distro? Maybe you have a problem with booting from USB sticks in general and the issul is not related to Guix System. <database2>how to install pci and zlib development library <ennoausberlin>database2: If its not in the store you need to write your own package definition <efraim>zlib is zlib from (gnu packages compression) <ennoausberlin>kmicu: I write the image to a sd card now and will check how many partitions are shown then <database2>ennoausberlin: i am trying to compile coreboot utils <efraim>there's a libpciaccess in (gnu packages xorg) and a pciutils in (gnu packages pciutils) <database2>will they work fine for installing coreboot utils <efraim>I assume so, I normally work by creating package definitions rather than compiling by hand <ennoausberlin>kmicu: Guess what happens. I wrote the guix image to /dev/sdb and now I can see the partition 1 to 4 with sdb3 is sudo blkid /dev/sdb3 <kmicu>ennoausberlin: just booted that iso on a UEFI laptop. No issues. <ennoausberlin>/dev/sdb3: LABEL="GUIXSD_IMAGE" TYPE="hfsplus" PARTLABEL="HFSPLUS" PARTUUID="21c6be38-4e91-4e60-a91b-84e71f367b5c" <kmicu>Did you disable secure boot? <kmicu>Then have no idea. For sure that very image works and boots on a UEFI laptop. <ennoausberlin>kmicu: If you run gdisk or fdisk on the stick. How many partitions do you see? <nckx>ennoausberlin, kmicu: The partition tables of those images are various hacks on top of each other (e.g. the image ‘is’ both an .iso and GPT-partitioned at the same time). Don't take them to mean much. <nckx>(gdisk will report a ‘damaged’ GPT table, for instance, this doesn't mean your download was corrupted.) <ennoausberlin>nckx: But even when I want to try to boot the image with an existing grub I get stuck during the mount step while booting the kernel <ennoausberlin>nckx: Kernel boots but while all the /dev/sd... stuff happens I get an error first and then a freeze <kmicu>ennoausberlin: I would find a working distro image and then we could compare it with Guix System iso. Everything boots here correctly so I cannot help much if I cannot reproduce the issue. <kmicu>Isn’t that hfs* stuff some Apple thing? <ennoausberlin>kmicu: Yes thats the old mac file system, but here I am using linux only <nckx>ennoausberlin: Eh? Who wrote that? <kmicu>ennoausberlin: what's your working distro? We could then take a look at its installation image. Maybe your hardware requires some workarounds. <ennoausberlin>nckx: The pastebin? Its a copy of my Ubuntu grub menu, because the stick did not boot at startup I used grub to boot the guix kernel this way <kmicu>Ah, so it’s a regression. Do you remember which GuixSD version booted correctly? <ennoausberlin>kmicu: Debian on some servers, too. But there are no usb ports on the debian machines <ennoausberlin>kmicu: 0.12 I think. But I dont remember how I wrote the iso at the time <kmicu>Biasecting ISO changes sounds like fun 😹 <ennoausberlin>My main goal was to study some scheme, so I decided to install a distro centered on scheme. But man, this is a hard path <nckx>There's a /boot/grub/grub.cfg on that hfsplus partition. Maybe try copying/booting from that. <kmicu>Please keep in mind it works w/o issues for more common hardware. <nckx>ennoausberlin: The /dev/sda3 in your pasted link looks suspicious, no? <ennoausberlin>nckx: Yes I took the data from that partition to create the ubuntu grub menu entry <nckx>ennoausberlin: Are you sure /dev/sda3 is where *Guix's* Linux kernel will see the installer once it boots? <ennoausberlin>nckx: Why suspicious? I used uuid before but it was not working anyway <nckx>ennoausberlin: If you're booting from Ubuntu's grub there's at least one other drive in the machine, which is at least as likely to be sda. <nckx>ennoausberlin: I think so, yes. <kmicu>Generally Guix System should manage GRUB entries. Chainloading from other distro is not supported. <nckx>kmicu: Tested perhaps, but it should work. <kmicu>nckx: What happens after reconfigure and reboot? <nckx>ennoausberlin: Use the label from the /oot/grub/grub.cfg on that hfsplus partition. <nckx>kmicu: We're talking about the installer, so I'm not sure what you mean. <ennoausberlin>kmicu: You dont know what I tried before :) I would like to see booting the stick from my UEFI-Bios. Thats for granted :/ <nckx>And it's certainly possible to kickstart a Guix system from a foreign GRUB. <kmicu>nckx: How is Ubuntu’s GRUB related to Guix System installer? <nckx>kmicu: Are we following the same conversation? Sorry for the confusion if not. <nckx>kmicu: ‘Its a copy of my Ubuntu grub menu, because the stick did not boot at startup I used grub to boot the guix kernel this way’ above. <nckx>And this should work if one sets all the dials correctly. <ennoausberlin>BTW: I already had lots of fun on another PC with 32Bit UEFI loader 64bit CPU and 64Bit Distro. And they told us UEFI is more flexible and less error prone. HA <kmicu>Ah, I wanted to solve booting and you are after a workaround to boot iso from an existing Ubuntu installation. Roger. <kmicu>Any idea why we see HFS+ there? <kmicu>ennoausberlin: do you have Ubuntu’s ISO version so I could compare it with Guix ISO to figure out why one boots for you and the latter not? <kmicu>nckx: did you have some code in mind in Guix repo? I don’t see anything related to HFS in it. <nckx>kmicu: I'm after both :-p <nckx>kmicu: Yeah, I'm looking around myself. <ennoausberlin>The original iso was Ubuntu 18.04 but later updated to 18.10 <ennoausberlin>OK. I will be offline for a while because of reboot and testing. Hope I can tell you good news using a label for root filesystem <phenoble>I'm looking for a new status-/tray-/taskbar for my system. I'm not familiar with new developments - does guix have anything recent packaged? <kmicu>Good luck ennoausberlin but if we point to wrong thing in the first place it dosen’t matter whether we use /dev/, UUID or label ;) <kmicu>Hi phenoble: xmobare, polybar, i3status, etc are available. polybar is probably the youngest one. <kmicu>ennoausberlin: maybe as nckx suggests that /dev/sda3 points to wrong thing so no matter UUID, lable, device path if will not work. 🤷 <ennoausberlin>kmicu: I replaced the /dev/sda3 with root=label=GUIXSD_IMAGE and updated grub. <nckx>ennoausberlin: Any reason for uppercasing it all? We use GuixSD_image. <kmicu>Probably cuz a tool returned that ‘/dev/sdb3: LABEL="GUIXSD_IMAGE" TYPE="hfsplus" PARTLABEL="HFSPLUS" PARTUUID="21c6be38-4e91-4e60-a91b-84e71f367b5c"’. <kmicu>(At least that HFSPLUS is hilarious so we can have some laughs.) <nckx>Will ignore that then. Use GuixSD_image instead. <nckx>kmicu: Why hilarious? It's just a regular HFS+ partition/file system. <nckx>You can mount -t hfsplus it if you don't believe me :-) <ennoausberlin>nckx: Hmm, the label is written by the dd writing process of the guix installer image and its shown so with lsblk afterwards <nckx>ennoausberlin: It *is* the label of the hfsplus partition, but we're not interested in that. <ennoausberlin>But wait, all partitions have the GUIXSD_IMAGE label. That can possible not work <nckx>Please try search --set --label GuixSD_image. <nckx>(Without the trailing .) <kmicu>nckx: why we use hfsplus in GNU Linux world to boot a GNU Linux dustro? Is this to support booting on Apple hardware? <nckx>kmicu: I think Danny is responsible (in a good way) for all the black magic. You'll have to ask them. *kmicu is sorry for all topos but migration to Colemak is still not completed xD <ennoausberlin>nckx: And the --root argument of the linux line needs to be updated how? *nckx switched from Colemak to Dvorak last year and feels your pain. <nckx>ennoausberlin: Good point. Try ‘--root=GuixSD_image’ too. <ennoausberlin>I will try it and if it works and you can explain, why it works I send you some beer :) *nckx is from Belgium :-p <kmicu>It lookes like dannym has only one commit in vm.scm and majority is u-boot related. <kmicu>My bad, forgot to check commits older than a year xD <nckx>kmicu: Might've been Marius. I've known to have been wrong before, but don't tell anyone. <kmicu>600 commits from dannym. Guix contributors are amazing. *nckx was sure they remembered a thread about all this but can't find it now. <kmicu>mbakke: ʕノ•ᴥ•ʔノ ︵ λ𝛌𝚲𝝀 Thank you. <nckx>I'm also quite puzzled how (and why) we have a fully mountable hfsplus partition on that thing without a single mention of it in the code base, kmicu. <nckx>Judging by the contents I'd say it's created entirely by GRUB. But it does have a /gnu/store directory structure. If that's pointing into ext4 file data… wow. ‘Nice.’ <kmicu>nckx: so that hfs thing is reported by Ubuntu’s tools not Guix? <nckx>kmicu: I run only the one true Guix System. <kmicu>And it reports HFS for you too? <nckx>It's a legit hfsplus party. ‘sudo mount /dev/loop0p3 -t hfsplus /mnt’ works. <nckx>ls /mnt → boot/ etc/ gnu/ mach_kernel mnt/ run/ System/ var/ <nckx>I don't know anything about OS X, but /System looks it. My guess is still GRUB being awesome. ***Piece_Maker is now known as Acou_Bass
<nckx>Damn, I forgot to tell ennoausberlin to remove ‘insmod hfsplus’ since that can only confuse GRUB. <nckx>sneek: later tell ennoausberlin: I forgot to tell you to remove ‘insmod hfsplus’ since that can only confuse GRUB. <kmicu>Maybe there’s some logic in thin madness “HFS… it can also be found on read-only media such as CD-ROMs.”. <kmicu>xdm 1.1.12 “This release collects 7 years worth of bug fixes and code cleanups, including the fix for CVE-2013-2179”. Still better than SLiM 😹 <kmicu>(Above is related to the discussion about default display manager for Guix System and why we should avoid SLiM as that choice.) <janneke>kmicu: maybe dannym's score is so low, because he is porting mes to arm; some 70 commits in mes and mescc-tools <kmicu>janneke: I’m not even implying that the number is low. I’m super thankful for those contributions and was surprised to see so many commits when I was searching for info about HFSPLUS in Guix System ISO. <kmicu>dannym’s score is so high is beyond expectations ヽ(*＾▽＾)／ <nckx>kmicu: Nitpick: it's not a score. :-p <nckx>Every commit is sacred ♪ <kmicu>You made the point even better nckx. Thank you. <nly>Currently, how do we build guile for a new platform? I see in package guile `(self-native-input? #t)', an included guile binary might be big? This is for ARM. <nly>I think janneke's (blog?) might have some info <ConfusedLizzard>Hey, short time user here. I decided to build my first package. Can anyone guide me to a place where i can look up, if the package is up to the guidelines? <pkill9>and just reading other package defintiions <ConfusedLizzard>If a package uses a lib that is under the agpl3, is it even allowed to release under the MIT license? <rekado_>ConfusedLizzard: the resulting combined work would be under the AGPL3. <rekado_>ConfusedLizzard: also check “guix lint your-package” <ConfusedLizzard>Is it even worth porting / publishing my package if the license is broken? <ConfusedLizzard>Is there any other file where the MIT license could be? Or is the MIT evil? <kmicu>ConfusedLizzard: there is no such thing as ‘MIT’. Search for ‘expat’. <kmicu>(‘MIT’ term is ambiguous like ‘cloud’. ‘MIT license’ doesn’t point to a specific license text.) <ConfusedLizzard>kmicu, wow this blows my mind. I always thought this was a specific license (with very few terms). Thank you <kmicu>(Not many folks pay attention to licenses and that Expat castero using AGPL-ed python-mpv as dependency is an example.) <rvgn>@kmicu What were those symbols? LOL, it did not render on my side. <kmicu>rvgn: aproxmating in ASCII something like (^_^)/ <g_bor>It seems to me that inkscap likes to use cairo at runtime. <g_bor>This is hard to notice, as it is usually available. <g_bor>Should we do something about that? <g_bor>It broke embedded pngs when ran in a guix environment -C <janneke>g_bor: let's make friends on a gtk/gnome conference ;-P <emyles>How do I connect to a port of a container started like this: guix system container my-config.scm <emyles>It starts a webdav service on a port, but how to connect from the host? ***rvgn-net is now known as rvgn
<emyles>Actually the guix command makes a script that is then run like this: sudo /gnu/store/h5...r8xf-run-container -net user <phenoble>So I keep finding packages for which there are definitions in the guix repo, but are not available on my up-to-date system(s) (- just done another round of guix pull && guix package -u guix). What gives? <g_bor>emyles: I believe we are still missing this feature. <g_bor>phenoble: Do you have guix in your profile? Are you on guix system? <phenoble>g_bor: um, so, I just did a "guix package -i guix", which does something. I guess that answers that first question. <phenoble>g_bot: No, i'm on two systems at the moment, debian- and ubuntu-based. <phenoble>Is there any way how to get the latest guix? Is this some stable/nightly/dev-kind of release-model being employed here? <emyles>Oh, thanks. I'll have to run my webdav using 'guix environment' or 'guix vm'. <rvgn>I have added my nextcloud account in the online accounts section of gnome system settings. But the online folder/drive is not appearing in the gnome file manager (Nautilus). What should I do? <ConfusedLizzard>Does anyone have experience with pythons setup.py build process? A lib required for a particular wont compile. <ConfusedLizzard>The build logs yield "Running "python setup.py" with command "build" and parameters ()" ...... "error: package directory 'generator/generate' does not exist" <ConfusedLizzard>While there is no generate directory, generator/generate.py exists and is mentioned in setup.py <ngz>I'm trying to update Wireshark. It's almost done but building fails during `validate-runpath' phase (log is at <https://paste.debian.net/1071269/>). Apparently, "/gnu/store/...wireshark/lib/" is missing from RUNPATH even though CMake build system is expected to add it. I assume it is removed somehow due to some configuration in CMakeLists.txt. <ngz>I tried to re-create RUNPATH, as done in "powertab" package, to no avail (all tests fail in this case...). <ngz>Help is more than welcome :) <ConfusedLizzard>emyles, thank you. Atleast i am not alone. If i check the logs/ content of the repo, i can gather that the files that should be generated are already inside the repository. The python-build-system call setup.py build which tries to generate but somehow fails. <ConfusedLizzard>I checked out other builds of python programs, which all use some kind of magic <rekado_>phenoble: you should not do “guix package -i guix” because you’ll always get an older Guix this way. Use “guix pull”. <rekado_>“guix pull” installs a new Guix to ~/.config/guix/current. Use that. <g_bor>phenoble: I believe the correct way to ahead from there, is to guix package -r guix, so it's not in your profile, then guix pull, maybe twice, as around 0.16.0 there was a change in how guix pull is handled. <g_bor>rekado_: should we do anything about this inkscape invisibly depends on cairo stuff? <phenoble>Using the binary from ~/.config does what I expect. <ngz>I'm trying to update Wireshark. It's almost done but building fails during `validate-runpath' phase (log is at <https://paste.debian.net/1071269/>). Apparently, "/gnu/store/...wireshark/lib/" is missing from RUNPATH even though CMake build system is expected to add it. I assume it is removed somehow due to some configuration in CMakeLists.txt. Does that ring a bell? <CornBurglar>Hey, I can't seem to find the different source options when defining guix packages. I know url fetch, git fetch, but I can't seem to find a "take from this directory" or any other options in the documentation, maybe I'm missing it? Maybe it doesn't exist? <emyles>CornBurglar: I don't know about docs but this works: <emyles>(source (origin (method url-fetch) (uri "file:///home/user/path/to/dir"))) <emyles>sneek: later tell CornBurglar I don't know about docs but this works: (source (origin (method url-fetch) (uri <emyles> "file:///home/user/path/to/dir"))) <emyles>,sneek: later tell CornBurglar I don't know about docs but this works: (source (origin (method url-fetch) (uri "file:///home/user/path/to/dir"))) <emyles>sneek: later tell CornBurglar: I don't know about docs but this works: (source (origin (method url-fetch) (uri "file:///home/user/path/to/dir"))) <nckx>sneek_: no botsnack for you. <mikadoZero> I have looked at the man page of `guix lint` and the manual section `Invoking guix lint`. How would I run `guix lint` on a package I wrote emacs-ace-link in emacs-xyz.scm? <mikadoZero>I am in the directory with the modified version of emacs-xyz.scm. These invocations did not work: `guix lint emacs-xyz.scm`, `guix lint emacs-xyz`, `guix lint emacs-ace-link`. <ngz>mikadoZero: Try ./pre-inst-env guix lint emacs-ace-link from Guix source tree.