The #guix survey also showed that our users are knowledgeable #Linux people. Almost 50% are experts/advanced and 47% are intermediate!
A lot of the attraction is the #nix value of declarative configuration and reproducibility. Users also identified that Scheme, Guile and Lisp are cool! Perhaps the overlap of two different communities! That was my path from #clojure and #ubuntu /#debian
See Q1 in the post:
https://guix.gnu.org/en/blog/2025/guix-user-and-contributor-survey-2024-the-results-part-1/
[#]scheme #emacs #guile #lisp #declarative #reproducible
=> More informations about this toot | More toots from futurile@mastodon.social
@futurile
"The GNU should be trying to meet users where they are to help liberate them, instead of creating an alternate reality where user needs are not addressed. This is a non-starter in the year 2024."
Don't know who wrote that, but that pretty much sums up the biggest meta-problem. :sadlinux:
It's got great ideas, but you gotta meet users where they are.
=> More informations about this toot | More toots from csepp@merveilles.town
@csepp Everyone's aligned on the desire for 'user freedom'. It's a question of the different routes people put forward that get's "us" to the end-goal.
The survey shows - people passionately love free software, #guix and they want to use it
Often in passionate groups we lose sight of the shared goal. Land-up discussing what divides, rather than what's shared!
For controversial topics - I hope the survey will let us think about "what can be done?" rather than "X is the only way"!
=> More informations about this toot | More toots from futurile@mastodon.social
@csepp i think that this is more born from market trends preferring products that present an easy to consume but "on the rails" experience.
Guix presents a well packaged scheme api and cli.
All the "simple" interfaces, eventually approach a general purpose control language as they improve, why not just start with one?
The learning process can be made informative, direct manipulation of the program syntax ( instead of a sequence of characters) could be a more preferable dev experience in general
=> More informations about this toot | More toots from mimlop@mastodon.social
@csepp especially with lisp. fire up a gtk gui, have gui elements for all the base data types and then youre mostly done with your code editor, now just analysis to aid in the editing process.
at runtime the user is just constructing sexps that are stored in memory.
=> More informations about this toot | More toots from mimlop@mastodon.social
@mimlop The problem is not with the choice of language, it's with the general priorities. Guile is an okay implementation language, G-expressions for example are very powerful. But Guile's implementation doesn't have the quality of life improvements that many mainstream languages have.
Not having an LSP implementation is maybe the biggest issue. Lisp advocates talk about S-expressions as if they magically solved refactoring forever, but I recommend actually trying some of the professional refactoring tools, like those in Jetbrains IDEs, then compare them with what you would have to do in Emacs to do the same.
Anyway, the much bigger issues are the contribution process and the packaging priorities. All the rest (docs, installer, bugs, etc) are just questions of time, but the arcane and archaic contribution process and the hard line free software and reproducibility stance is a question of leadership priorities. Luckily there seems to be some (although not substantial) interest by the core devs to improve on the former, but not so much the latter.
And sadly that overshadows all the really fricking cool technical stuff for me. It can have G-expressions for all I care, it's not going to fix the issue that packages are shipped with known vulnerabilities for months or are just completely broken for a year.
=> More informations about this toot | More toots from csepp@merveilles.town
@futurile Something that is sorely missing from the survey evaluation (so far, no idea what's coming in part 2) is a consideration of confounding factors, like who the survey would even reach and would be motivated enough to fill it. The growth stat for example only makes sense if we assume that those who left are willing to put in the effort into filling out the essay. My guess is that those who just bounced off after a quick test are not going to be represented.
=> More informations about this toot | More toots from csepp@merveilles.town
@csepp I may not be fully understanding what you're saying ...
Yes, a limitation of ALL surveys is you only get information from people committed enough to fill it out.
Other avenues like analytics and user research aren't possible in FOSS.
That feels like common sense to me! I resisted caveating:
"this is imperfect information our knowledge is partial" - because it doesn't get you anywhere.
We had no data, now we have some - lets use it to improve Guix, no?
Am I missing your point?
=> More informations about this toot | More toots from futurile@mastodon.social
@futurile I would love to reach out to the beginner/explorer type, for good.
After all, that’s the whole point of the “practical user freedom” stance, which led to the overall design and even to the way some of the documentation is written.
Probably we need more in the way of popular education?
=> More informations about this toot | More toots from civodul@toot.aquilenet.fr
@civodul Things that would help from the data:
=> More informations about this toot | More toots from futurile@mastodon.social
@civodul
From the graph, it seems that the install process imposes a clear cutoff here.
Consider that the survey includes a strong bias: we only have a return from people answering it. Those below the cutoff probably consider not useful providing any feedback from their fail experience.
@futurile
=> More informations about this toot | More toots from csantosb@sigmoid.social
@csantosb @futurile @civodul « Those below the cutoff probably consider not useful providing any feedback from their fail experience. »
Yes, we are observing the bias itself. 😀
That’s said, to reach out to the “Beginner/Explorer“ profile, we have two directions, IMHO:
• Make Guix “easy”; @futurile describes some 9 of many items which could help.
• Make Guix “simple”, it’s probably more in the spirit “user freedom”; something à la “Linux From Scratch“.
Personally, I am more aligned with “simple” than with “easy”.
=> More informations about this toot | More toots from zimoun@sciences.re
@zimoun @csantosb @civodul Yeah it's a good point - there's definitely different definitions and paths. Plus points for referencing one of my favourite talks ;-)
=> More informations about this toot | More toots from futurile@mastodon.social
@zimoun
In my experience, #guix is already simple for most users. Documentation is easy to follow and clear enough to keep going.
My point is that the cutoff originates during the install itself, a game over for not technically skilled people.
@futurile @civodul
=> More informations about this toot | More toots from csantosb@sigmoid.social
@csantosb Your analysis assumes that the “Beginner/Explorer” profile only jumps in installing Guix System bare metal, when VM and/or Guix on Foreign Distro are also options. Well, somehow your intuition’s probably right about the “why” of this cutoff, I don’t know.
My point’s we cannot say much because we’ve not enough data to explore this bias itself so I refrain about this “why”. 😀
Instead, we observe the “Beginner/Explorer” profile did not answered, therefore:
Morevoer, the documentation’s on “simple” side but not “easy” side, IMHO.
@civodul @futurile
=> More informations about this toot | More toots from zimoun@sciences.re
@zimoun
I don't even consider Guix System when it comes to not skilled people, way too complicated.
Only foreign, as a first step towards using Guix.
@civodul @futurile
=> More informations about this toot | More toots from csantosb@sigmoid.social
@futurile @civodul @zimoun
An example of my whole point here: 1.5 hours to just install guix.
https://logs.guix.gnu.org/guix/2025-01-29.log#130603
Most users just give up after such first experience, and so the cutoff in previous graph.
=> More informations about this toot | More toots from csantosb@sigmoid.social
@csantosb I agree that 1.5 hours for the first “guix pull“ is a bad user experience.
However, give up after this experience isn’t related to the “How knowledgeable a Linux are you? (Q1)”. Is it?
I mean, I consider myself as « Intermediate (e.g. comfortable with the command-line and configuring many aspects) » or « Advanced (e.g. you correct the Arch Linux Wiki!) » and I would probably give up myself with such user experience.
@civodul @futurile
=> More informations about this toot | More toots from zimoun@sciences.re
@zimoun @futurile @civodul Don’t get me wrong: I just wanted to point out that, based on missing feedback from the survey corresponding to beginners/explorers, probably due to failure to install guix, provides an interesting insight about why some people don’t use guix, and so, on how to to improve the situation.
=> More informations about this toot | More toots from csantosb@sigmoid.social This content has been proxied by September (3851b).Proxy Information
text/gemini
