IRC channel logs

<kevinfish>How do I get paste working under i3 in icecat?

<mb[m]1>kevinfish: Are you referring to preserving newlines? Unfortunately I don't have an answer, but I'm surprised about the assertion that it's i3-specific (I have the same problem on i3). Did you verify that it worked under say GNOME?

<kevinfish>no, when I tried to install a normal desktop from boot it got too complicated, so I went with the minimal GUI install. When I try and paste, either with a C-v or by cording the mouse buttons it just jumps the browser into a very strange place. Doesn't seem to work at all. Like if I'm on facebook and I have a field I'm trying to paste into with lots of comments in a sub-thread it jumps to a place where the main posting is all th

<kevinfish>..backwards go get back where I was. If I click "comment" to enter a comment all my previous typing is just gone

<kevinfish>Also it GuixSD seems to be VERY slow for the class of machine I'm using, config files I set in my non-privileged user account get overwritten on boot, and important utilities (like "mount") are no longer available in the root user

<wigust>kevinfish: Config files for user overwritten on boot? Do you mean user home directory? Guix doesn't touch them at all as any distro.

<mb[m]1>kevinfish: Oh, that seems to be a different bug. What kind of machine are you using?

<mb[m]1>Also, as wigust mentioned, GuixSD only touches files managed by your system configuration.

<wigust>kevinfish: If you need to you could check an example for XFCE and GNOME sessions https://git.savannah.gnu.org/cgit/guix.git/tree/gnu/system/examples/desktop.tmpl

<kevinfish>yes, the files config files in my home directory don't seem to be staying. Also I've noticed applications don't seem to be staying installed. Is there a cli pastebin utility?

<mb[m]1>You can use http://sprunge.us/ to "pipe" to a pastebin.

<wigust>kevinfish: Also https://paste.debian.net

<brendyn>kevinfish: Does C-y paste in icecat?

<kevinfish>well, I can't copy/paste to a web interface. Chicken or the egg situation

<kevinfish>nope

<brendyn>i was using i3 before and never had issues

<wigust>kevinfish: Why? Right mouse click and paste

<mb[m]1>kevinfish: Doing multiple `guix package` invocations at once may cause the new profile to get "overwritten", but existing profiles are immutable.

<kevinfish>well, that worked here. Cording the mouse works here too. Maybe its a JS issue on facebook.

<wigust>kevinfish: Icecat doesn't allow proprietary JS by default. https://www.gnu.org/philosophy/javascript-trap.html

<kevinfish>I must have a seriously munged up system here. When I sudo su into root I don't even have ls. I do echo <dirname>/* on the dirnames in my path and only /bin/sh is there

<kevinfish>I take it that's not normal right?

<wigust>kevinfish: Try "sudo -i"

<kevinfish>ahh, much better!

<mb[m]1>`sudo su -`would work too; they both perform a "normal" login, causing /etc/profile to be sourced (which sets up essential variables such as PATH).

<brendyn>kevinfish: i had that same issue. just run `su' then run `source /etc/profile'

<kevinfish>got it. Its going to take a bit of getting used to this

<brendyn>kevinfish: then, at least in my case, you neet to edit /root/.bash_profile

<brendyn>my /root/.bash_profile is if [ -f ~/.bashrc ]; then . ~/.bashrc; fi

<brendyn>and then my /root/.bashrc is the default one installed in guixsd, which loads /etc/profile

<kevinfish>ok, thanks

<mb[m]1>brendyn: That has been the default ~/.bash_profile in GuixSD for many years, also for root.

<bms_>Hello.

<brendyn>Would somebody mind testing ecryptfs? I get an error saying it refers to /bin/mount

<brendyn>run `mkdir a b; sudo mount -t ecryptfs a b'

<brendyn>mb[m]1: Yes I just found that I must have overridden mine when I was copying files over to my new system

<bms_>Hello.

<janneke>morning!

<efraim>hello!

<efraim>i accidently enabled some zooming feature in XFCE on guixsd, any idea how to turn it back off?

<efraim>alt + scrollwheel

<bms_>Hello.

<bms_>ACTION is back (gone for 00:10.07)

<brendyn>Has anyone got gnome-settings-daemon running with a tiling wm like i3 on guixsd?

<mb[m]1>Holy, the python.scm split-up coincided poorly with the python-updates merge to core-updates.

<mb[m]1>ACTION attempts the merge

<rekado>:(

<rekado>sorry about that!

<brendyn>Anyone know how to setup automatic harddrive mounting? and getting them recognised in pcmanfm etc?

<mb[m]1>Being able to grep the commit log for file names and variables is a life-saver.

<mb[m]1>rekado: No worries, it had to be done.

<rekado>brendyn: I needed to add gvfs to the system profile to get one-click mounting in nautilus.

<rekado>mb[m]1: I also relied heavily on the uniform commit logs to move copyright assignments along with the package definitions.

<brendyn>Also, I get all these weird devices like pts, chm, store, blikio appear in nautilus/pcmanfm

<brendyn>in i3. they never appeared in GNOME3

<brendyn>rekado: I already have gvfs.

<brendyn>brb rebooting

<rekado>I think we can build approach gradle by building groovy 2.0.0-beta3

<rekado>it still comes with an ant recipe that shows how to bootstrap groovy

<roptat_>rekado: groovy is part of maven dependencies, so I have something

<rekado_>roptat_: which version have you been working on?

<rekado_>groovy requires gradle and gradle requires itself, so I wanted to build an older groovy first

<rekado_>and then build gradle with just groovy

<roptat_>I use 2.4.10 (the latest when I built it)

<roptat_>it requires gradle for its build system, but it can build fine with our ant-build-system

<roptat_>although I only have a minimal version that's enough to run one small script in another package

<roptat_>part of groovy is written in groovy :/

<mb[m]1>Merged!

<mb[m]1>efraim: Can you try to build python-pytest on the latest core-updates when you're available?

<roptat_>I only built the subset that's written in java

<roptat_>you can get my definition from here: https://framagit.org/tyreunom/guix-more/blob/master/more/packages/java.scm

<efraim>mb[m]1: sure

<efraim>hmm, did audacity pass -msse before?

<rekado_>roptat_: the build.xml that comes with 2.0.0-beta3 includes a bootstrapping step.

<rekado_>I think that’s to build a minimal groovy compiler from Java sources first and then use that on the rest of the sources.

<rekado_>I think it may be better to use that build.xml instead of the one the ant-build-system generates for us.

<roptat_>yay... probably

<roptat_>I didn't look older versions of groovy too closely because they all required a groovy compiler, so I figured it would be just like maven: better to build the latest version and ignore previous ones

<roptat_>but if it does provide a bootsrapping step that doesn't require groovy, it's worth trying to build 2.0.0-beta3

<mb[m]1>efraim: I have a bunch of updates queued for core-updates that will invalidate all your substitutes. Building them now and will probably push later/tomorrow.

<efraim>mb[m]1: ok

<efraim>python-pytest built for me on core-updates

<mb[m]1>efraim: Yay! Thanks for verifying. That means the merge was successful.

<efraim>is nettle part of that?

<efraim>just got the email from gnu-announce

<mb[m]1>Nope, haven't gotten to that. Will you take it?

<efraim>i'll try to get to it

<mb[m]1>I have libxml and libxslt, and a bunch of miscellaneous x11/"staging" updates.

<mb[m]1>Building Guile takes ages.

<rekado_>takes around 45mins on that 1.5TB RAM server at work

<rekado_>I guess having 192 cores is useful sometimes.

<efraim>I expect more along 4 hours on my firefly 3399

<efraim>trying to browse nettle's source on gitlab with links was incredibly frustrating

<jonsger[m]>rekado_: If I would have 1.5TB RAM at work would be nice :)

<efraim>nettle built without any problems

<efraim>apparently audacity indirectly depends on nss

<castilma>hey, guix system reconfigure tells me using device in bootloader configuration is deprecated and I should use target instead. 1. the doc should be updated to mention this. 2. does it mean the `target` of mapped devices ($ info guix mapped).

<castilma>s/.$/?

<castilma>what if i dont use luks nor raid(the only types for mapped devices acording to the manual)?

<roptat_>castilma: this is what I use: (bootloader (grub-configuration (target "/dev/sdc")))

<roptat_>so just replace device with target here

<roptat_>I think the change happened after 0.13, so it's not yet in the online manual

<g_bor>Hello guix!

<g_bor>I've just checked the hydra status on giux, and we have this failing build called tarball.

<g_bor> http://hydra.gnu.org/build/2199154

<g_bor>From the log it seems, that convert is missing, ehich I guess shoul be in imagemagick.

<castilma>roptat: thanks. but afaik im using the doc from the git checkout.

<rekado_>g_bor: hi!

<vagrantc>so, i've done a few installs now, two with a locally built installer, and i'm confused why the installer consistantly installs a kernel that's older than the one the boot media uses

<vagrantc>(e.g. it's running 4.14, but the installer is installing 4.13.x to the newly installed system)

<mb[m]1>efraim: icu4c fails to build for me on 'core-updates'.

<efraim>i'll try it out

<efraim>it built on aarch64

<efraim>how is it failing?

<mb[m]1>Test failure. Will investigate with -K.

<vagrantc>is there any commandline flag to get guix system init to be more verbose? the most recent build of it i have just seems run indefinitely at 125% cpu utilization with no output

<vagrantc>ACTION let it run for 8 hours with no progress

<rekado_>sounds like you have a loop in your configuration somewhere.

<wigust>vagrantc: --verbosity=[from 0 to 5]

<vagrantc>heh. --verbose just would be too easy.

<mb[m]1>efraim: It wasn't a test failure (was watching intermixed build output), it's actually a compile error.

<mb[m]1>"number_decimalquantity.cpp:333:14: error: expected primary-expression before ‘float’"

<vagrantc>seems to be looping on: downgrading to read lock on `/var/guix/temproots/4549' ... acquiring write lock on `/var/guix/temproots/4549'

<vagrantc>just when i think i have the hang of this...

<castilma>any tips on how to dualboot guixsd and nixos? would you recommend chainloading? both create many menuentries. I guess it is not possible for one system to recognize the system options the other wants, right?

<castilma>s/system options/grub boot entries/

<g_bor>Hello guix!

<g_bor>I've just run a make check on a fresh git checkout, and I have the guix-system test fail.

<g_bor>Where does that system definition come from?

<castilma>g_bor: I experienced the same and sent it to bug-guix.

<wigust>g_bor: I assume it's not "operation system definition", but a "guix system" thing.

<wigust>Oh, wrong read, sorry.

<janneke>g_bor: is this part of: make check-system?

<janneke>system tests are in gnu/test/*

<g_bor>ok, I have found it...

<g_bor>it dies on the qemu module one...

<janneke>are you on GuixSD?

<g_bor>yes

<janneke>hmm, on foreign distro's i've seen kvm/qemu setup quirks -- not on GuixSD

<g_bor>it seems, that EXIT trap is executed...

<g_bor>when the test.sh is run from another location it executes and exits 0.

<mb[m]1>I suspect the ICU issue is due to the incomplete fix for this: https://ssl.icu-project.org/trac/ticket/13385

<mb[m]1>Odd that it doesn't trigger on Aarch64 though. Maybe i18n is not enabled on that platform?

<mb[m]1>Aaah, actually the problem seems to be GCC, specifically this: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=14608

<mb[m]1>When both <cmath> and math.h is included, GCC fails to locate std::signbit.

<castilma>i installed guixsd on my machine and it boots but there are 1.5 problems: 1. the boot hangs: lsh tries to generate a key, but it only gets 31 bits of randomness, it asks me to give it more by typing random stuff, but: 1.5 no keys seem to get registered. unplugging and replugging it results in kernel infos being printed on the console, saying it recognised a new device. any ideas? maybe this gets better if it booted until the end: can i

<castilma>vide guixsd/lsh more randomness through a file?

<dogetest>is there something like package group?

<g_bor>bye

<mb[m]1>castilma: For lsh, if you have console access, "mashing" the keyboard does work IIRC. Although I would recommend using OpenSSH since lsh is more-or-less unmaintained.

<castilma>mb[m]1: the thing is i am stuck on the console and can't input anything. i typed about 100 keys. is the input maybe blind? i will try moving my fist over the keyboard.

<mb[m]1>castilma: I think you might need more than that. Another option is to "ping-flood" the machine from another host, if you have network.

<alezost>dogetest: instead of groups we use packages. "lxde" or "gnome" are the examples of such "group" packages

<mb[m]1>efraim: Did you try reproducing the icu4c build failure? Maybe it broke with glibc 2.26?

<efraim>it just builds fine here, maybe there are optimizations for other platforms that aren't there for aarch64?

<mb[m]1>Interesting.

<mb[m]1>I think this is the actual bug: https://sourceware.org/bugzilla/show_bug.cgi?id=22296

<mb[m]1>I'll try to pick the corresponding commit.

<bms_>Hello.

IRC channel logs

2017-11-19.log