<rekado_>(and I do think it should be part of Guile)
<civodul>but in the end, if you prefer your version htgoebel1, that's fine too :-)
<htgoebel1>rekado: I strongly disagree with "->" being clearer - at least for guile beginners.
<civodul>i think the conclusion is that it's the kind of tool that should be available in Guile, indeed
<htgoebel1>civodul: Okay, then I'll keep mine :-) Thanks
<civodul>htgoebel1: there's a precedent though, "->" exists in Clojure, and while clarity is in the eye of the beholder :-), i do think that it's fairly easy to "get" what it does
<htgoebel1>civodul: If "-Y" would be defined elsewhere, it would be okay for me. But since it needs to be defined here by a syntax-rule, makes it hard to understand. One needs to understand the syntax-rule first.
<cbaines>It's not actually shown, but I am storing some translations for lint warnings, so I'm hoping to build on this to add more internationalisation
<cbaines>I also haven't included lint warnings in the comparison between revisions, but I'm interested in that for reviewing patches, as it's hopefully really useful to see if any lint warnings are being introduced
<xavierm02[m]>Alright. I found out why guix import crate doesn't work for some packages: in guix/guix/import, things are extracted from the JSON with an and-let*, and because of this, when there is no homepage, the script doesn't work.
<gnu_srs1>roptat: This time bash crashed... Took a screen shot of the qemu console. I have to replace all four: bash mkdir tar xz and put them into the .xz file too.
<xavierm02[m]>I don't really understand the idea behind the and-let*. Especially since it fails with something that looks like a network error. And also, if you just replace the and-let* by a let*, it works and magically finds the home-page anyway :o
<rekado_>gnu_srs1: start “qemu” with “-curses” in a terminal
<rekado_>xavierm02[m]: and-let* is used to abort when a necessary previous step failed, so that worse errors or unnecessary work is avoided.
<Minall>pkill9: For example, a reconfigure when I only have gnome, and basics utilies installed with the same versions as in debian
<Minall>JUst that would be cool, like for example, since debian is so stable since it goes with 'old' versions of the packages, If I want to work on a 'stable debian' environment, I would install the same packages with the SAME versions as in Debian
<Minall>And the same kernel of course, but in linux-libre
<pkill9>yes that would be possible, if you put in the time to write the package definitions
<Minall>Yes, And I can have another reconfigure for the latest packages, and another for example, with packages to work with something exactly
<Minall>Also... How can I add emacswm to a reconfigure?
<pkill9>you wouldnt need to use reconfigure, you could put the package definitions in another channel, and create a manifest with the debian packages that you want, and use with `guix environment`
<xavierm02[m]>rekado_ I understand that. But why should a missing home-package field in the JSON returned by crate lead to a failure at all? The behaviour for the license (i.e. setting it to #f) seems fine.
<GNUtoo>My use case is the following: I'm evaluating GuiX to see if it can be deployed in production, so we could step by step move the Replicant infrastructure to Guix, when new things get suported in GuiX
<g_bor[m]>GNUtoo: yes,you can manually configure using ip, and it will be lost on reconfigure.
<GNUtoo>Here my goal is to fullfil the most basic use cases like using a basic web server and such, and I'll probably get a public ipv4 for that
<GNUtoo>That would work, but it'd be better if there was a quick and dirty workaround for that (like issue commands in the scm or generate or copy some config files that would at the end make sure that the interface has an ipv6)
<GNUtoo>(it'd make testing less manual and less error prone, and at the end of the day the hack will be gone when deployed)