all messages for Guix-related lists mirrored at yhetil.org
 help / color / mirror / code / Atom feed
From: Julien Lepiller <julien@lepiller.eu>
To: guix-devel@gnu.org
Subject: Re: Thoughts and questions from a newcomer
Date: Tue, 12 Sep 2017 09:59:33 +0200	[thread overview]
Message-ID: <9283C7CD-7C71-4624-B125-7DD135CEDDDF@lepiller.eu> (raw)
In-Reply-To: <877ex47r7h.fsf@localhost.i-did-not-set--mail-host-address--so-tickle-me>

[-- Attachment #1: Type: text/plain, Size: 3516 bytes --]

Le 12 septembre 2017 01:25:54 GMT+02:00, Andrew Erlanger <andrew.erlanger@gmail.com> a écrit :
>Hello, Guix users,
>
>After a bit of love, I've got most of my personal computing environment
>running in GuixSD. As I get more comfortable with the system, I hope to
>naturally become a developer.
>
>At this point, I have a few thoughts and questions, which I'm looking
>for some feedback on, especially where I'm misguided.
>
>It seems that the main problem Guix is trying to address is about state
>and side-effects. In modern computers, a program running over here
>can affect the execution of a completely unrelated program running over
>there because of side-effects. The result is a large amount of
>unnecessary complexity. Effectively, the programs have many, many
>inputs, and with feedback loops between components, the system can
>exhibit chaotic behaivor.
>
>If I'm not mistaken, Guix attempts to provide a way of building
>programs so that all inputs to the build process are known exactly.
>Assuming that the build is a deterministic process (which I
>hope it is!), this ensures that the resuling executables are not
>tainted by side-effects.
>
>The main thing that has been confusing me about this process is in
>dealing with configuration files. Clearly the state of a program such
>as, for example, wpa_supplicant, depends on the configuration file used
>to start the process. So, if we are trying to effectively declare the
>state-machine of wpa_supplicant using Guile, shouldn't the config file
>be part of the Guile definition of wpa_supplicant?
>
>My main draw to Guix was my frustration with configuring Gentoo and
>other distros. After I spend a day exactly configuring a system, I want
>the state which I set up to be _exactly_ reproducible anytime I want.
>But clearly I don't need all the information of the entire disk image
>to reproduce that exact state; I just need a few configuration files
>worth of information.
>
>If I can't include the contents of configuration files as an input to a
>package build, then I can't really capture the state-machine of the
>package with a single declaration. I have to then use something like
>stow and git to manage my config, which is not really solving the
>problem.
>
>On the other hand, it seems that Guix might be more about mitigating
>side-effects rather than declaring state. So, which is it?
>
>Thanks for reading,
>
>Andrew

Hi Andrew !

We want to be able to distribute packages to people, so we can't make a package depend on some configuration file. As you noted, it would be unreproducible.

On the other hand, when using guixSD, you can use system services to declare your configuration. Our wpa-supplicant-service cannot be run alone, but if we had a better service, you could declare something like this in guile:

(service wpa-supplicant-service-type
  (wpa-supplicant-configuration
    (ssid "my-wifi")
   ...))

This would create a configuration file in the store and create a new herd service for wpa_supplicant that would use this file. You may be interested in the system configuration section of the manual if you want to learn more.

So, you have a reproducible package and a reproducible config, therefore a reproducible behaviour.

With guixSD you have only one file to edit in order to configure your system entirely. At least that's part of what we would like to achieve. One of your first contribution could be this wpa-supplicant-service-configuration :).

[-- Attachment #2: Type: text/html, Size: 4005 bytes --]

  reply	other threads:[~2017-09-12  7:59 UTC|newest]

Thread overview: 3+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2017-09-11 23:25 Thoughts and questions from a newcomer Andrew Erlanger
2017-09-12  7:59 ` Julien Lepiller [this message]
2017-09-17 19:32 ` Ludovic Courtès

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=9283C7CD-7C71-4624-B125-7DD135CEDDDF@lepiller.eu \
    --to=julien@lepiller.eu \
    --cc=guix-devel@gnu.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
Code repositories for project(s) associated with this external index

	https://git.savannah.gnu.org/cgit/guix.git

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.