IRC channel logs
2021-09-12.log
back to list of logs
<paulusASol>ArneBab: maybe download an older, fitting version from snapshot.debian.org? <ArneBab>paulusASol: I now installed Mercurial as source, installed python3-dulwich, and got hg-git. That works. Trying to install git manually failed, because it did not find openssl. <gnu_srs>youpi: I see that procps is high up at: graph-total-top.txt. There is a bug report with patches at: #987557 since 25 April 2021. <gnu_srs>Unfortunately the maintainer does not update the package since 6 April. Maybe you can build a +hurd package for unreleased? <ArneBab>Does anyone test the Guix VM? I tried to install there but the install was extremely slow and ultimately failed, so I went back to the debian images. <paulusASol>The problem is, that git and mercurial FTBFS on the hru <ArneBab>ah … is Mercurial configured with rust-support? <paulusASol>so the architecture-specific packages (git, mercurial) and the arch-all packages (git-man, mercurial-common) get out of sync <ArneBab>I just compiled Mercurial on the Hurd manually — worked flawlessly <akib>do you mean guix on hurd? i tried it using pre built qemu images from guix ci servers but i discovered two problems <akib>first, there was no network, not even localhost. however it was possibly my fault, passing some extra args to qemu solved it <akib>second, herd can't connect to shepherd socket after vm reboot. so i have to run fresh images every time i want to use it <akib`>ArneBab: rust... is it free? hyperbola is going to switch to bsd, and rust is one of the causes. they say rust users don't have freedom 3 because of trademark problems <ThinkT510>wanting to protect a trademark (in effect a brandname) is very different to to equating something as non-free <akib>is there any free rebranded version of rust? <ThinkT510>for what purpose? everyone is working on rust itself <ThinkT510>maintaining a programming language can be a massive undertaking. why fork and rebrand it? you'll forever be playing catchup <ThinkT510>yes you can fork it (which in my book makes it free) but you just can't call the fork rust (if you change and redistribute it) <akib>that means all programs written in rust are non-free <ThinkT510>If I write a program in rust and licence it as ISC then the program is free <ThinkT510>I could even licence it as GPL, which is also free <ThinkT510>writing a program in rust is totally different to forking the entire rust programming language <akib>but the build dependency -- the rust compiler -- is non-free <ThinkT510>I don't understand the quibble. It seems like a complete non-issue to me. The naming of something is totally separate to its function and the availability of the source. <youpi>gnu_srs: about procps, we already have a +hurd version <akib>ThinkT510: from what i understand, rust users don't have freedom 3, that mean a free system can't have rust, and that means a free system can't compile those free GPL licensed rust programs <akib>ThinkT510: source available software isn't always free <akib>can someone say me why kernel hackers dislike (and sometimes hate) c++? <ThinkT510>SerenityOS devs certainly don't share that view about C++ <akib>trademarks are the problem. they forbid releasing modified versions of rust without their explicit approval <akib>some hackers use c++ (including me, but i don't know whether i qualify as a hacker) <gnu_srs>youpi: Seems like the +hurd version is 2:3.3.16-5+hurd.1, while the patches in #987557 is for 2:3.3.17-5. <youpi>I mean, what is the actual problem with the current package? <youpi>I don't plan to upload each and every patched version, unless there is an actual benefit <gnu_srs>If there are no new build-dependencies on 2:3.3.17-5 compared to 2:3.3.16-5 of course building a hurd version is not interesting. <jrtc27>and take it elsewhere if you really want to persist <akib>jrtc27: but changing the name all over the source code is a huge task. see <https://www.gnu.org/distros/free-system-distribution-guidelines.en.html>. it says "Trademarks are associated with some software. For example, the name of a program may be trademarked, or its interface may display a trademarked logo. Often, the use of these marks will be controlled in some way; in particular, developers are commonly asked to remove references <akib>to the trademark from the software when they modify it. In extreme cases, these restrictions may effectively render the program nonfree. It is unfair for someone to ask you to remove a trademark from modified code if that trademark is scattered all throughout the original source." <akib>ok, rust is off topic here. where should these philosophical discussions go? <nckx>Ideally nowhere. There's nothing unfree about Rust. Claims to the contrary are false, not philosophy.