IRC channel logs


back to list of logs

*civodul sends message about IPFS support
<aldenp>question; is there an equivalent to Portage's USE flags in Guix?
<aldenp>I'm a Gentoo user thinking about trying out Guix, but I can't live without mah USE flags
<Formbi>on the guix site someone wrote that there is if I'm not mistaken
<buenouanq>Is the USE flag thing a legit desire, or just Gentoomind incapable of thinking about things any other way?
<buenouanq>I didn't mean for that to come out as rude as it maybe did...
<aldenp>no offense taken; I just want a way to control what features are and aren't built into the package without manually screwing with the package's build script and changing the dependency list
<buenouanq>My question then is, is there a `Guix way' to do that already? might not look anything like whatever you're use to, but would have similar results
<buenouanq>or is the USE flag approach a real and important enough thing that there should be some kind of direct equivalent
***atw` is now known as atw
<THFKA4>sounds like it would go directly against the advantages of reproducible builds and binary caching
<tune>you could spin up your own recipe that builds the relevant package with different settings probably
<pkill9>there would have to be cache for a lot of different vriations on packages
<brendyyn>Is it possible to emulate a system where the clock runs faster inside the emulator? like `sleep 5s` would actually sleep for 1s inside the emulator?
<reepca>It's possible for different outputs of the same derivation to refer to each other in a cyclic manner, right?
<reepca>e.g. "out" depends on "doc", "doc" depends on "out"
<rekado>reepca: I don’t think that’s permitted.
<g_bor>hello guix!
<reepca>rekado: just checked my store database, seems so indeed, though I'm curious how that's enforced. Out of curiosity, what's the reason behind that restriction? It's fine for store items to refer to themselves, after all.
<reepca>("seems so indeed" referring to the /not/ being permitted, just realized that may be ambiguous)
<rk4>hello mr g_bor
<reepca>ah, seems it's enforced by just refusing to register anything with cycles
<jlicht>hey guix!
<roptat>hi guix!
<roptat>I'm having some troubles with my file manager under guixSD
<roptat>it lists a lot of useless file systems (pts, shm, store, …)
<roptat>and it doesn't list usefull ones like my usb or sd device
<roptat>The solution was to use dbus-launch in
<jlicht>roptat: Does this work with any windows manager? I've had this problem for many weeks as well
<pkill9>I have that problem as well in Thunar
<jlicht>FYI, I am using exwm atm, so I am not sure .xsession is even used
<jlicht>roptat: How does your .xsession look to fix this issue?
<brendyyn>roptat: I reported this bug ages ago but couldn't fix it
<brendyyn>Actually Danny was helping me with it bug I ended up travelling for a month so I didnt continue
<brendyyn>So yeah, it would be great if someone more knowledgable than me could figure it out'
<pmikkelsen>hi guix, it's been a while :) Is there no way to see the IRC log anymore?
<demotri>pmikkelsen: See the topic, there is a new log:
<pmikkelsen>demotri: ah, thanks
<demotri>pmikkelsen: You are welcome :-)
<g_bor>hello guix!
<g_bor>I've came here for some insights about the format of jmod files in java 9 and above
<g_bor>it seems that they are not simple zip files, as unzip complains, that there are 4 extra bytes at the beginnig.
<g_bor>any idea?
<g_bor>from my digging it seems so far, that the first for bytes are J M 0x01 0x00. Then a standard zip file follows.
<g_bor>I assume the second part is some version number.
<roptat>g_bor, or a file format?
<roptat>if anyone is interested, my .xsession looks like "exec dbus-launch openbox-session"
<roptat>g_bor, I've been working on improving my parser today
<roptat>I found a bug that created an infinite recursion when parsing multiline strings, and I fixed a few things in the antlr grammar again
<roptat>I also found that the context-free grammar in the documentation of scala doesn't match one of the files of the compiler itself ^^'
<roptat>I also worked on the typer, and I have the very simple parts of it (I can now determine the full name of explicit types, but I still don't resolve implicit types)
<pkill9>in this example of 'package-input-rewriting', where would i put the name I want to rewrite the package to?: (package-input-rewriting `((,openssl . ,libressl))))
<pkill9>the docs say "Optionally, rewrite-name is a one-argument procedure that takes the name of a package and returns its new name after rewrite."
<pkill9>but I can't work out where to put it
<efraim>one option I guess would be 'set!', but wouldn't make use of package-input-rewriting
<efraim>'git grep' shows two examples in (gnu packages guile)
<pkill9>ah thanks a lot
<efraim>'grep -C5 package-input-rewriting doc/guix.texi' is actually a good example
<g_bor>roptat: that is nice :)
<g_bor>( the work you have done, not that the file does not match the grammar )
<roptat>g_bor, thanks
<pkill9>I pass it a string but it errors with 'wrong type to apply'
<g_bor>roptat: now I believe that I will do the following with jmod-s: check that the first 4 bytes are matching this magic,
<efraim>gnuastro@0.8 built fine on aarch64, testing on armhf now
<efraim>and I might have a fix for vim-full with skipping test87
<g_bor>then strip these 4 bytes off, repack them as they were zips, then prepend the four bytes back
<efraim>and a version bump
<efraim>nvm, dependencies of gnuastro don't build on armhf
<roptat>pkill9, if the doc says it's a procedure, it can't be a string I think :)
<roptat>g_bor, sounds like a plan
<g_bor>:) and let's hope that they will change the magic when this no longer works :)
<pkill9>how do i pass it a procedure and what do i pas sit?
<pkill9>pass it*
<roptat>pkill9, no idea, sorry :/
<pkill9>ok thanks anyway
<roptat>looking at the example in guile.scm, it's a procedure that takes a string and returns a strin
<roptat>it's used to change guile-* to guile2.0-* for instance
<pkill9>yeah i saw that, i tried to use "package-name" as the procedure but it still said it's the wrong type
<pkill9>unless i gave it to it incorrectly
<pkill9>oh, i think i know hwat's happening, i'm using the whole thing incorrectly
<pkill9>hmm i'll see
<g_bor>pkill9: I've had a look at that, it seems that the procedure when used should map a string to a string.
<g_bor>package-name is not in that form.
<g_bor>I believe that the default value (identity) is what you really want, given that you specified package-name
<pkill9>so what would I put there instead of a string?
<g_bor>pkill9: what exactly are you trying to achieve?
<g_bor>If the identity mapping of the old package name to the new is ok for you, then you can simply omit that
<pkill9>to change the name of the package that I get from the package-input-rewriting
<roptat>you should define a procedure, for instance (package-input-rewriting `(...) (lambda (name) (if (equal? name "foo") "bar" name)))
<roptat>that renames "foo" to "bar", but anything else is kept as is
<roptat>see how guile-variant-package-name returns a procedure? I think that's what you missed
<roptat>calling guile-variant-package-name once returns that procedure (it's like a partial function call)
<g_bor>pkill9: yes, just as roptat wrote, if you want to do this unconditionally, then you can just (lambda (_) ("newname")), but this will rename all packages passed to simply newname...
<pkill9>i'm only using it for one package though, so it will only rename that package?
<pkill9>hmm I used that example you gave g_bor and I still get "wrong type to apply" when I run `guix package -A newname`
<g_bor>oops, I messed up my parents.
<g_bor>it should be
<roptat>(const "newname") actually
<g_bor>(lambda (_) "newname")
<roptat>(which is the same as what g_bor said)
<g_bor>yes, const looks nicer :)
<pkill9>thanks it accepts it now
<pkill9>but when i try to build it it starts building boost
<pkill9>maybe this thing is for more complex stuff i don't understand
<quiliro>hello...i think i am almost ready to make a package on my own...i made one in cooperation with rekado. But now i want to try by myself. I will start with MICMAC, a 3D reconstruction software. It is professional grade.
<quiliro>it has just a few depends that i think guix already defined and builds with cmake and make install
<buenouanq>ok, I gave in and preordered the Librem5
<buenouanq>When will I be able to install GuixSD on it?
<rekado>quiliro: excellent! Let us know if you need any help with this.