IRC channel logs
2023-02-05.log
back to list of logs
<gabber>where are differences in package definitions determined? <nckx>gabber: A different way to think about this that might help is you explicitly asked Guix to build something called wayland 1.21.0 using the sources of wayland 1.20.0. By mistake, but you still did. You gave it a URL as *hint* (not identifier) where it might find them, but since that URL is bogus Guix can't tell the difference between a broken upstream and a mistaken hash. <a12l>If you're going to add a "guix file" to a repo containing a guix package description, what would you name it? <mirai>unmatched-paren: indentation is off for the last string-append? <mirai>other than that, I don't see any obvious deficiencies <unmatched-paren>mirai: that's an attempt to fit it into the horizontal width limit of ~80 characters <attila_lendvai>strugging with debbugs: how do i merge two bugs? 60002 should be closed and merged into 36069. send a control message to the one that should remain open: 'merge 60002'? <mroh>just merge should be enough. And be patience, it can take several minutes. <attila_lendvai>ACTION has sent the merge from emacs and waits with fingers crossed <attila_lendvai>no, i spoke too early: it has closed the one i wanted to keep open: 36069 (reopening it now) <Guest46>I am using virt-manager to create a VM that has TPM 2.0 but I get "qemu-system-x86_64 does not support TPM backend type emulator". How can I fix that? <bjc>try installing ‘tpm2-tss’ <Guest46>Done already, still the same message. <bjc>iirc you can give libvirt the path to the tpm emulator somehow <bjc>this is on a manjaro host, though, not guix <Guest46>I have installed ovmf package as well but I can't choose other than bios. Is it because I am not using Guix System? <Guest46>Ah I know why, it requires configuration, though I don't know how. <lechner>attila_lendvai / unmatched-paren / mroh / Hi, 'forcemerge' is usually easier to use than merge because it resolves status inconsistencies between the bugs that might otherwise prevent a merge. also, closing either of the two bugs after the merge will close both (and is hence a mistake). when it is finally time to close the bug, it it is better to close via email to XXX-done, because closing bugs requires a reason, which cannot be <sneek>Welcome back lechner, you have 1 message! <sneek>lechner, haugh says: A while ago, I saw you were looking at composing forms around abort on #f. Have you seen chain-and from SRFI-197? The syntax-case implementation works for me; you just need eof-object from (ice-9 binary-ports) and let-values from (srfi srfi-11) <xd1le>Hi! I just booted into a newly installed guix system \o/ But how do I shutdown? <xd1le>I know it's halt, but how do I shutdown immediately? <bjc>halt should be immediate; at least it is on my computer <xd1le>I guess for me it hangs, or maybe it's taking a while, shrug <xd1le>Guest46, I'll try that, thanks. I just force shutdown by holding power button and now I gtg <mirai> (keyboard-layout keyboard-layout))) <mirai>how does the last keyboard-layout work? <mirai>the first one is a field from the xorg-configuration record type <mirai>I suppose it "refers" to the field from operating-system but where/how does that work <mirai>but I'd like to know how exactly this second one works <apteryx>mirai: see the implementation of define-records* in (guix records) <mirai>apteryx: can the field-names be used as accessors of the same record (provided its not something self referencial like (field field)? <mirai>maybe accessors is not the right word <ss2>mirai: Thanks for your work in refactoring mpd's service. Amazing! I just tried to set up the service, but unfortunately there's no output. <mirai>ss2: there's a small bug in it <mirai>this will need to be speedily merged to fix the endpoint field <ss2>I was just about to ask. :) <mirai>ss2: in the case you're not using the endpoints field, could you share your configuration? <ss2>Nothing special so far. <mirai>did you erase the user / group fields? <ss2>no, but I was having odd messages there if I'd use another user. <mirai>I presume you're setting the user/group to your "interactive" user <ss2>I was planing to use the services default user. <ss2>On a side not, I've noticed that the service is missing gnutls to authenticate https streams. That has been the case for the old service before. It'd be good to have that lib included too. <mirai>I'm unsure about environment-variables default value, I left it as empty but for pulseaudio users, it will need to be set to (environment-variables <mirai> (list "PULSE_CLIENTCONFIG=/etc/pulse/client.conf" <mirai> "PULSE_CONFIG=/etc/pulse/daemon.conf")) <mirai>note that this might not do what you expect <mirai>I've no idea how pulseaudio interacts between different users <mirai>you could test with and without these env variables and have an interactive user session <ss2>the mpd is noninteractive, right? I'm hearing that term the first time. <mirai>see if system mpd user pulseaudio interferes with the interactive one <mirai>mpd is noninteractive/system user, yes <apteryx>I was wondering; if I use the static-networking-service-type, can I omit specifying routes and name-servers? Would it use some sane defaults for these then? <lechner>apteryx / i don't think so. nameservers are the domain of DHCP, for one <apteryx>I see; indeed in (gnu services base) for <static-networking> we see the routes and name servers default to '() (none) <apteryx>it also has "FIXME: doesn't belong here" about static-networking-name-servers ^^ <mirai>I don't think it will pick any defaults for those because "nothing" is also a sane value <mirai>a default route isn't required <mirai>nor are nameservers (you can tweak nscd to just use hosts or mdns) <apteryx>Yeah I'm used to DHCP where the name-server is auto-detected <apteryx>Do I need a default route to have network connectivity to the Internet when using a the static network service? <lechner>actually, that's probably automatic as long as you declare the gateway <lfam>mirai: For 60357 (MPD missing inputs), does this change enable some features that people are missing? Or is it just for the sake of completeness? <mirai>I guess so? unless you want to manually insert the routing table of the whole internet <mirai>lfam: enables features that are missing <mirai>there are some more features that are still missing but I didn't dig much further into them <ss2>mirai: I'll wait for your small fix to be pushed. <mirai>I've no upnp setup to test if it will really work <lfam>And for 61792, can you give a brief explanation of the bug that is being fixed? Like, what goes wrong without this patch? <mirai>lfam: its a ungexp of a list <lfam>If you can't test UPNP, and nobody else is asking for it, I think it's fine to leave it disabled <mirai>which won't work because after ungexp, the list gets turned into a procedure call <mirai>which is nonsense ("string" "more strings" ...) <lfam>mirai: Would you say that 61297 is a followup to "services: mpd: Refactor MPD service."? <dcunit3d>i'm using emacs-* packages from guix on a foreign distro, but whenever i use `C-h k` or `C-h f` to look things up, i get an error for "(void variable read-symbol-positions-list)". it's pointing to the correct files in the hashed /gnu/store package AFAIK. is there something I need to do to set this up? <lechner>cwebber / Hi, thanks for those links last week! What was your motivation (or use case) for syrup over straight preserves, please? <lechner>i see the ordering and like it, but am otherwise unsure of the benefits above canonicalization <lechner>Hi, why does guix shell -f guix.scm not ensure that the build product exists, please? I wish it would invoke guix build -f guix.scm if it didn't <Guest46>How can I configure `virt-manager`with `ovmf`? <mekeor[m]><dcunit3d> "i'm using emacs-* packages..." <- does it work with emacs -Q? <dcunit3d>i'm not loading any emacs-* packages without my init though <dcunit3d>after running via `emacs -q` and manually loading packages from my config, i'm seeing the problem <dcunit3d>it also doesn't load help from dependencies loaded from straight, same error <mirai>what compile flags does guix use? <lfam>What do you mean, mirai? <mirai>lfam: does guix set things like -fomit-frame-pointer? <mirai>the CFLAGS used in general for guix packages <unmatched-paren>mirai: look at guix/build/<SOMETHING>-build-system.scm to figure that out <unmatched-paren>mirai: there's no default CFLAGS for gnu-build-system at least, it seems <lechner>it may just be -g -O2, plus platform flags <mfg[m]>Hi, where does the wrap-executable' phase of cmake-build-system' come from? i can't find it in guix/build-system/cmake.scm. <unmatched-paren>mfg[m]: build-system doesn't contain the code that runs at build time :) <unmatched-paren>rather, it tells guix how to turn a package with that build system into a builder .scm script <unmatched-paren>this script will refer to the guix/build/*.scm files for the default phases <mfg[m]>Alright, thanks unmatched-paren :) <nckx>lechner: Very minor feature request: a list of URLs that make no sense to <title>, like paste.debian.net. <nckx>unmatched-paren is correct, there's no forced ‘-O2 -fwhatever’ for Guix packages (Guix uses this flag to build *itself*, nothing more). However, autotools does default to -O2 -g, so it's quite common. But no, no distro-wide -funroll-loops -mplacebo. <mfg[m]>how do i know the actual label of an input? i'm looking at the build failure of itk-snap and the problem seems to be that gnu/packages/image-processing.scm:1470 doesn't find a suitable labeled input and theregore returns #f which leads to a build failure <nckx>unmatched-paren, attila_lendvai: ‘merge 36069 60002 \n close 60002’: ‘merge’ makes bugs synonyms of each other, so combining it with ‘close’ makes little sense (and will in fact close both if the merge succeeds—since they are now 2 names for 1 bug). <nckx>Sorry, just replying to random old backlog… :) <nckx>mfg[m]: If you're using the ‘new’ label-less input style, my recommendation is not to try to guess the label. Whatever it is, it's not API, and could change. Instead, use helpers like search-input-file. If that's not an option, it's fine to keep the ‘old’ style. There are other packages that do just that. <nckx>mfg[m]: …*however*, my guess is that here it's "qtdeclarative", no "-5". <mekeor[m]><unmatched-paren> "could somebody review #61214..." <- heh, i could, but i'm a little fatigue after travelling so much. i need a little rest. but i bet it's very good and you don't really need my feedback :D <paulapatience><mekeor[m]> "crossposting from reddit: https:..." <- The running emacs tangle property isn't set to main <neox>Good evening. I'm trying to update a python lib that depends on gtk+ and I obtain an error during the check phase : "RuntimeError: Gtk couldn't be initialized. Use Gtk.init_check() if you want to handle this case" <neox>I made the lib depend on gtk+ and python-pygobject <neox>Ok, I found out what xorg-server-for-tests is x) <mekeor[m]><paulapatience> "That name is begging for guix/..." <- yes, i agree, its a bad name, also difficult to distinguish qua pronounciation <mekeor[m]><paulapatience> "Also, "she script" -> "the..." <- i should have said i'm not the author of gix <mirai>??? am I supposed to add nss-certs as inputs to a package? <mirai>perl SSL connections fail if I don <mirai>don't add it as a package input <paulapatience>> <@mekeor:matrix.org> > <@paulapatience:matrix.org> Also, "she script" -> "the script" <paulapatience>I wasn't sure, and I figured (correctly) that you'd clarify :) <paulapatience>> <@mekeor:matrix.org> > <@paulapatience:matrix.org> That name is begging for guix/gix typos <paulapatience>> yes, i agree, its a bad name, also difficult to distinguish qua pronounciation <paulapatience>The author's last name is Gile; this looks like a pun on Guile/Gile/Guix/Gix <mirai>that's not what I'm asking though <mirai>since almost no package pulls nss-certs <mirai>but for some odd reasons, a package dependent on perl-io-socket-ssl refuses to budge unless nss-certs is a package input <mirai>`guix shell nss-certs <package-in-question>' doesn't work <attila_lendvai>ACTION has not much clue about perl, just needed that package for something else