IRC channel logs

2018-04-26.log

back to list of logs

<nlcted>What's the policy on packages creating hidden directories in home; is that permitted in general, if it's generally necessary?
<OriansJ>nlcted: well generally the only hidden directories are part of the XDG spec https://specifications.freedesktop.org/basedir-spec/basedir-spec-latest.html
<OriansJ> https://wiki.archlinux.org/index.php/XDG_Base_Directory_support
<OriansJ>Basically all well behaved programs generally conform to that standard, does that answer your question nlcted ?
<nlcted>I've noticed that some packages, like TeXlive, make their own hidden directories.
<nlcted>So, it's not generally allowed, then?
<bavier>nlcted: it's not disallowed, and would be prohibitively difficult to "fix" in so many packages
<OriansJ>nlcted: it is more generally considered bad manners but we understand if it isn't possible and you need to create a hidden directory
<nlcted>Alright; I really appreciate the fast response; I hope to contribute to Guix soon.
<nlcted>Goodbye for now.
<baconicsynergy>any chance of xsltproc becoming a guix package?
<rekado>sneek: later tell baconicsynergy xsltproc is part of libxslt
<sneek>Will do.
<rekado>sneek: botsnack
<sneek>:)
<rekado>some web pages now use JP2000 for images, which cannot be displayed by epiphany.
<g_bor>hello guix!
<g_bor>I've seen that rekado posted two emails in connection with GSoC, but I did not see the same communication about the third GSoC project, related to Shepherd. Does that communication belong to an other mailing list?
<efraim_>I believe it should all be on guix-devel
<efraim_>rekado: I changed the second overdrive to port 52522
<rekado>sneek: later tell g_bor I haven’t sent out the third email yet.
<sneek>Got it.
<rekado>efraim_: yes, all communication should be on guix-devel. These emails are to the individual students to encourage them to introduce themselves on guix-devel and conduct all project-related communication on guix-devel if possible.
<snape>OriansJ: "freedesktop.org is open source / open discussion software projects working on interoperability and shared technology for X Window System desktops. The most famous X desktops are GNOME and KDE, but developers working on any Linux/UNIX GUI technology are welcome to participate" It seems pretty restrictive, I can't understand how their standard can be universal. Plus, I'm not sure what they mean by "Linux/UNIX". And anyway
<snape>they state it later: "freedesktop.org is not a formal standards organization".
<civodul>Hello Guix!
<efraim_>civodul: I changed the port for the second overdrive
<zybell_>snape:why dont you read these both sentences in the other order? Then they are easier to understand. Translated from third person standardese to first person normal:"We are no formal standards organization,our max power is to recommend" The (now second)sentence:"Our field of endeavour is XWindow Desktop Environments,especially KDE and GNOME,although other Linux or even UNIX Projects are invited to participate." If one does know the history
<zybell_>too,many assumptions in these sentences read even more clearly.
<snape>zybell_: you're changing the original text, which I don't think is fair. See https://www.freedesktop.org/wiki/.
<zybell_>I translated from standardese to meaning. which I announced and think very fair. And short enough for IRC to boot.
<OriansJ>snape: yes that may be true but in regards to what standards we have in regards to hidden directories. I can't think of any other standard that would apply. Hence an informal standard is better than no standard
<OriansJ>zybell_: translating, I believe is best described as a group activity for finding collective meaning. However we collectively agree what a standard means is via our collective action.
<zybell_>Fair enough if you include in collective action actions by lawyers and such in case of *formal* standards;-)
<civodul>efraim_: re the overdrive, thanks!
<efraim_>Np
<OriansJ>zybell_: I would never discount the contributions that lawyers can make as they are part of the collective assumptions we have created as part of civilization.
<civodul>efraim_: well it's seems picking this port number wasn't enough
<civodul>i'll have to let rekado investigate the firewall rules
<civodul>efraim_: besides, i've created accounts for rekado and myself, as well as the build farm user account
<civodul>BTW, 'sudo' is configured to ask for root's password
<civodul>do you know how to change it?
<rekado>so outgoing SSH to port 52522 is blocked?
<rekado>could you remind me of the IP?
<rekado>I’ll have to open a ticket with IT for that; can take a while.
<rekado>(has taken > 1 week in the past)
<snape>OriansJ: There is a HN discussion about it: https://news.ycombinator.com/item?id=4331855. It's not obvious which solution is the best, but I'd not favor one over another.
<rekado>efraim_: I opened a ticket to change the firewall settings.
<snape>s/but/so
<efraim_>civodul: I'll try to make it not ask for a password, should be a simple change to /etc/sudoers
<civodul>efraim_: i think it's ok to ask for a password, just not root's
<rekado>looks like “salmon” sometimes crashes at runtime; the most likely cause is that it was built with Boost 1.66 instead of 1.5x.
<rekado>I may have to add an older version of Boost.
<rekado>I just built tensorflow
<rekado>but the build system is weird.
<rekado>it just gave me a benchmark binary and a static library libtensorflow-core.a
<mbakke>rekado: Heh. Nice achievement regardless!
<bavier`>rekado: cool
<bavier`>rekado: tensorflow is mostly a library, right, so maybe not too weird?
<rekado>bavier`: yes, though I’d like to have a shared library and more than just the core set.
<rekado>it’s obvious that they want me to use bazel instead of cmake.
<bavier`>oh of course
<civodul>congrats on this first milestone, rekado
***civodul changes topic to 'GNU Guix | https://gnu.org/s/guix/ | Outreachy+GSoC: https://gnu.org/s/guix/blog/2018/guix-welcomes-outreachy-gsoc-and-guix-hpc-interns | videos: https://gnu.org/s/guix/blog/tags/talks/ | bugs: https://bugs.gnu.org/guix | patches: https://bugs.gnu.org/guix-patches | paste: https://paste.debian.net | log: https://gnunet.org/bot/log/guix | Guix in high-performance computing: https://hpc.gu'
***civodul changes topic to 'GNU Guix | https://gnu.org/s/guix/ | https://gnu.org/s/guix/blog/2018/guix-welcomes-outreachy-gsoc-and-guix-hpc-interns | videos: https://gnu.org/s/guix/blog/tags/talks/ | bugs: https://bugs.gnu.org/guix | patches: https://bugs.gnu.org/guix-patches | paste: https://paste.debian.net | log: https://gnunet.org/bot/log/guix | Guix in high-performance computing: https://hpc.guixsd.org'
<efraim_>i'm starting to think the way to fix the enlightenment setuid programs issue is to change those binaries to install to /run/setuid/ and then to manually install them to $out/....
<civodul>efraim: not sure what you mean
<civodul>the store cannot contain suid binaries
<efraim>i know
<civodul>bavier`, rekado: i just noticed that Singularity fails to execute the images that 'guix pack' currently produces due to a bug in their sanity checks: https://github.com/singularityware/singularity/issues/1487
<efraim>the build process installs the binaries to the store and then runs chmod and chown on them, which of course doesn't actually happen, then I've added them to /run/setuid-programs in my service
<efraim>but I keep getting errors that the backlight can't change or the cpu frequency doesn't auto change and things like that
<bavier`>civodul: interesting
<civodul>efraim: does E have /run/setuid-programs/ first in $PATH?
<civodul>that could be the reason
<efraim>I'm not sure how to check it's path, I know it added /gnu/store/...enlightenment/bin to the start of the path, but these binaries are in /gnu/store/..E/lib/..
<civodul>you can check /proc/N/environ
<rekado>Singularity has one interesting related project: the scientific file system https://sci-f.github.io/
<rekado>I think that it may be useful for Guix to provide a pack backend for this; it seems to be a modular alternative to the big Docker blobs.
<civodul>i watched the asciinema thing but didn't quite understand the story
<rekado>civodul: BTW, some of my colleagues are very excited about this unshare wrapper
<rekado>what “asciinema thing”?
<civodul>there's an asciinema "video" on the sci-f web site
<civodul>i'm glad people are excited about the 'unshare' wrapper :-)
<civodul>it sounds fun and *potentially* useful, but i always wonder how useful it is in practice
<rekado>it’s way easier to use than it is to install Docker.
<civodul>right, but the counter-argument would be "Docker is already installed"
<rekado>(it’s also easier than installing Guix on a “hostile” system)
<civodul>right
<formbi>hi
<rekado>Docker seems to be less popular now than even a year ago.
<pkill9>what's the 'unshare' wrapper?
<formbi>how to configure network on the GuixSD livecd?
<rekado>formbi: I suggest looking at the manual. It should be available on another VT.
<rekado>ACTION needs to go now
<formbi>found it, thanks
<civodul>roptat: another Scheme on Android: http://www.blogbyben.com/2018/04/lightweight-r7rs-compliant-and-in-palm.html :-)
<civodul>from http://scheme.dk/planet/
<civodul>ACTION -> []
<efraim>here is some of the relevant parts I think https://bpaste.net/show/15a5af92d49e , perhaps it's finding the binary by looking in the E_LIB_DIR for it
<efraim>although PANTS=ON isn't relevant to most things
<efraim>relevant code! src/bin/e_fm/e_fm_main_eeze.c: snprintf(buf, sizeof(buf), "%s/enlightenment/utils/enlightenment_sys", eina_prefix_lib_get(pfx));
<efraim>where if '%s' is E_LIB_DIR then it matches the path exactly
<dustyweb>ok so!
<dustyweb>I'm setting up my librem with guixsd (yay!)
<dustyweb>this time I'd like to do real actual full disk encryption, not just /home
<dustyweb>however... I'm not sure what to do. I guess I would have a single partition, but then I can't mark it as bootable
<dustyweb>and we can't do a separate /boot/ I think?
<dustyweb>since it needs to access /gnu/
<dustyweb>I guess I could have just /gnu/ in a separate partition?
<dustyweb>rekado: what did you do? other thoughts?
<mbakke>dustyweb: I encrypt the whole disk, GRUB can decrypt it these days; even /boot.
<dustyweb>mbakke: oh, yay then :)
<mbakke>I'm halfway through packaging "fastboot".
<janneke>ACTION compiled tcc using only mes+mescc -- no gcc/binutils/libc or guile used
<vagrantc>ACTION claps excitedly!
<mbakke>Thankfully Debian has already done the grunt work. The android repositories come with no build instructions, no Makefiles, and needs rearranged headers to work with the standard libraries.
<mbakke>janneke: Wow :) What is the next target?
<janneke>mbakke, vagrantc: thanks! ... next target is getting it built as a guix package
<janneke>there's a wip-bootstrap branch that has a "tcc-boot" package, building that still fails
<efraim>oh wow
<janneke>i'll be releasing mes-0.13 when tcc builds in guix together with some kind of roadmap
<nckx>This is awesome (in both senses of the word).
<janneke>some work needs to be done in the stage0-hex assembler->mes path
<janneke>much work needs to be done in the mes/mescc/tcc->gcc patch
<janneke>I'm working with OriansJ to get their amazing M2-Planet transpiler to build mes.c (or actually, currently to build the second simplistic scaffolding stage leading up to mes.c)
***uniq10_ is now known as uniq10
<dustyweb>ok mbakke, so...
<dustyweb>problem with putting it all under sda1
<dustyweb>"warning: this GPT partition label contains no BIOS Boot Partition: embedding won't be possible."
<dustyweb>but cfdisk wouldn't let me mark the giant sda1 as bootable under gpt
<dustyweb>thoughts?
<dustyweb>I think the librem is using coreboot, so
<dustyweb>oh I see
<dustyweb>there were instructions to leave a bios partition that I didn't follow ;P
<cbaines>The system tests that I've tried seem broken currently on the master branch... they get stuck in QEMU somehow
<cbaines>Is anyone else experiencing this?
<lfam>cbaines: Could be the recent changes to QEMU. Can you bisect?
<lfam>Also, I think this was reported on one of the MLs
<cbaines>I'm testing using an older commit at the moment, it that works, I'll try and narrow it down
<lfam>cbaines: See <https://bugs.gnu.org/31268>
<bavier`>cbaines, lfam: It may be related to the bug posted recently on help-guix
<bavier`>lfam: yup, that's the one
<lfam>Does anyone know off-hand why most of the distro depends on jbig2dec?
<cbaines>Ah yes, thanks for finding the problematic commit (37b9be5878d1694967a41d313de97f7a957df120)
<lfam>(The SVG I got from `guix graph --reverse-package` is humongous)
<lfam>Maybe it goes through ghostscript into cairo, GTK+, and balloons out from there
<lfam>It seems suboptimal that we have multiple PDF renderers with huge reverse dependency graphs
<pkill9>this could be a good alternative to chromium https://iridiumbrowser.de/
<bavier`>pkill9: cool
<bavier`>oh, it even removes chromium's phone-home antifeature
<bavier`>I think mbakke said they based some of their patches on iridium's
<civodul>efraim, rekado: the 2nd overdrive seems to be at work now \\o/