Ludovic Courtès

GNU, Guix, Guile, reproducible research, etc.

Public Key

npub15teuvnz55q44z0xlq33dh02x7das02u79dxpah6fmaj9c5g70nwsqdzz7y

NIP-05 Address

Profile Code

nprofile1qqs29u7xf322q2638n0sgckmh4r0x7c84w0zknq7maya7ezu2y08ehgpz4mhxue69uhhyetvv9ujuerfw36x7tnsw43qz9thwden5te0wfjkccte9ekk7um5wgh8qatzyhsgul

Publishing to

relay.ditto.pub

relay.mostr.pub

Author Public Key

npub15teuvnz55q44z0xlq33dh02x7das02u79dxpah6fmaj9c5g70nwsqdzz7y

Show more details

Last Notes

2026-06-16 10:05:51 UTC

I thought Java had long left me alone but I saw two backtraces over the last couple of days, including one from a printer, and it felt like a blast from the past.
The beans 🫘 are still alive!
https://toot.aquilenet.fr/system/media_attachments/files/116/759/212/226/860/414/original/f0e9930e444f1bab.png

2026-06-05 08:21:24 UTC

Just discovered “Principia Softwarica”:
https://principia-softwarica.org/
It’s inspiring both as a piece of work—one person learning and documenting an entire OS code base over several years!—and as a demonstration of what designing software with consistency, clarity, and minimalism in mind can produce.
(HT @nprofile…8z6r !)

2026-05-16 10:35:16 UTC

- reply

@nprofile…ygzq Also, I’m sure we can avoid insinuating things and being tongue-in-cheek and have an even more fruitful discussion!
@nprofile…23kd

2026-05-16 10:33:25 UTC

- reply

@nprofile…ygzq I still support Software Heritage in its mission to archive source code—I think it’s key to free software and to reproducible builds.
I’m well aware that they enabled Hugging Face’s LLM training (as well as other non-LLM-related experiments). I’m not a fan of that and of other attempts to follow the AI trend, but that doesn’t make Software Heritage an AI shop at this point.
@nprofile…23kd

2026-05-12 13:21:24 UTC

“Time travel without borders”
https://guix.gnu.org/en/blog/2026/time-travel-without-borders/
On the freshly implemented ability of ‘guix time-machine’ and ‘guix pull’ to download channel files without putting you at risk.
#Guix

2026-05-11 15:23:06 UTC

These tech companies and the authorities that enable them are insane. (And it’s just one example among hundreds worldwide.)

2026-05-07 20:22:28 UTC

- reply

@nprofile…8xwg It’s “just” a layer on top of LaTeX, Lout, etc., but maybe give Skribilo a try:
https://nongnu.org/skribilo/
And for music scores, LilyPond, of course. :-)
@nprofile…fl4q

2026-04-29 09:55:47 UTC

- reply

@nprofile…fl4q Nice! What characteristics are you referring to?
@nprofile…7dhr @nprofile…ygzq

2026-04-27 10:12:34 UTC

The mechanism that makes it so that we no longer need to enter our LUKS passphrase twice is pretty cool:
https://guix.gnu.org/manual/devel/en/html_node/Bootloader-Configuration.html#Automatic-LUKS-Master-Key-Passing
#Guix

2026-04-23 12:27:20 UTC

Well-documented list of free software projects and their use of genAI:
https://codeberg.org/small-hack/open-slopware
It’s already a long list that shows what looks like uncritical adoption, both by high-profile projects (systemd, VLC, etc.) and by niche projects (GNU Mach is a prime example).

2026-04-02 20:14:30 UTC

- reply

@nprofile…mwda C’est ça qu’est triste : de plus juste pouvoir ignorer le bouzin et continuer sa vie comme avant.
Tout ça parce que le bouzin est en train de bouffer tout ce qui se trouve autour de nous et que trop peu ont su/ont pu le tenir à distance.

2026-04-01 20:36:14 UTC

Watching the activity log of guix-cuirass-bot is interesting:
https://codeberg.org/guix-cuirass-bot?tab=activity
It makes ~20 comments per day.
#Guix

2026-04-01 14:54:03 UTC

- reply

@nprofile…rwr4 CONDA is way worse than today’s Debian (which has been leading the Reproducible Builds movement, has “buildinfo” files for provenance tracking, “rebuilders”, etc.).
That one can rebuild the same package with CONDA, as this page claim, or that there’s any sort of provenance tracking, is a lie.
I took a look at conda-forge in https://hpc.guix.info/blog/2021/09/whats-in-a-package/ (towards the end). Better than PyPI, but far from being transparent, let alone reproducible.

2026-03-31 21:24:24 UTC

- reply

@nprofile…ge3w Any writeup from Rocq folks about the things you’re referring to? I’m curious.

2026-03-31 15:48:26 UTC

- reply

I find it quite amazing that they’d make strong claims like this. Congrats for daring! :-)

2026-03-31 15:47:44 UTC

- reply

Thing is, this simply cannot be true.
First because CONDA relies on software provided by the host distro.
Second because dependency resolution gives different results at different times.
And there are probably other issues, such as lack of hermetic builds, but these two are the big ones.

2026-03-31 15:45:14 UTC

- reply

That led me to look at the CONDA web site, in particular this:
https://conda.org/blog/conda-practical-power/
It’s amazing how that page talks about #reproducibility and provenance tracking, going as far as saying that CONDA not only supports “traceability and rebuildability” but even lets you answer: “Where did this binary come from, and how was it built?”. No less.

2026-03-31 15:41:59 UTC

Great CONDA/Guix comparison by @nprofile…rwr4 today 👇
https://hpc.guix.info/events/2025-2026/caf%C3%A9-guix/

2026-03-30 13:49:18 UTC

Le ☕ Café #Guix de demain sera animé par @nprofile…rwr4 et comparera #Guix et #CONDA pour des projets scientifiques.
📅 mardi 31 mars, 13h–14h
▶ https://hpc.guix.info/events/2025-2026/caf%C3%A9-guix/
#RechercheReproductible #ScienceOuverte

2026-03-30 13:44:46 UTC

💡 Did you know? There’s an API to get data about builds etc. from Cuirass 👇
https://codeberg.org/guix/cuirass/issues/130#issuecomment-12381921
#Guix

2026-03-27 21:39:33 UTC

- reply

@nprofile…l5x9 A PC (Intel CPU) on a PCI board for use in an UltraSPARC workstation: https://en.wikipedia.org/wiki/SunPCi

2026-03-26 21:53:39 UTC

- reply

I’m amazed to see quite a few people knew about SunPCi—the Fediverse is a special place. :-)
I found it in a box I hadn’t opened in a while (I have the host as well, an Ultra 5, with a real CPU!). This SunPCi thing always stroke me as a funny hack from a hardware vendor that must have been desperate.
I don’t think I ever got it to work back when I played with this; it probably required some Solaris proprietary stuff.

2026-03-26 18:00:07 UTC

Hey you, #retrocomputing nerds! Any idea what this is?
https://toot.aquilenet.fr/system/media_attachments/files/116/296/575/811/221/255/original/1c65f20107ec320e.jpg

2026-03-26 14:56:14 UTC

When spending ~500 M€ of European public money on an #HPC cluster from a US company, use of said machine is subject not just to EU and national laws, but also to US export regulations and sanctions:
https://lumi-supercomputer.eu/termsandpolicies/
#sovereignty

2026-03-24 15:50:06 UTC

- reply

@nprofile…kwhx Overall it works pretty well (it’s quite amazing actually) but there are occasional issues, like https://codeberg.org/guix-science/guix-cran-scripts/issues/5

2026-03-24 13:29:19 UTC

If you’re into #Rstats and/or #bioinformatics, if you care about #ReproducibleResearch, consider helping out fellow #Guix hackers!
https://codeberg.org/guix-science/guix-cran-scripts/
The scripts above are what powers the automated imports of CRAN and Bioconductor as Guix channels, and they need ❤

2026-03-24 13:05:53 UTC

- reply

@nprofile…hdk0 We can have the doc at doc.guix.gnu.org too. :-)

2026-03-21 21:06:22 UTC

- reply

@nprofile…j0m9 Modules are loaded sequentially.

2026-03-21 18:25:17 UTC

- reply

But getting a self-contained, declarative system config makes it worth the effort!

2026-03-21 18:24:27 UTC

- reply

The main difficulty was that substitutes were missing for some of the cross-compiled packages (like Linux-Libre). And for PulseAudio (and tdb!) I had to struggle a bit to make package variants that can be cross-compiled.

2026-03-21 18:23:26 UTC

- reply

The good thing is that this finally led me to create a Guix System image to work on it, and it went well!
I wrote a derivative of the A20-OLinuXino-Lime2 image type already in Guix proper:
https://codeberg.org/guix/guix/src/branch/master/gnu/system/images/a20-olinuxino-lime2.scm
‘guix system image’ then cross-compiles everything.

2026-03-21 18:20:28 UTC

I have an Olimex A20-OLinuXino-Micro (ca. 2017) whose primary purpose is to run PulseAudio with an RTP source that I can send to from my laptop to play music (it’s connected to a tuner-amp from the 80’s—youngsters have no idea what I’m talking about :-)).
Thing is, during FOSDEM, there was a power outage and it never came back. It was running an old Debian with Guix atop (so I blamed systemd for that, without evidence).

2026-03-21 18:17:16 UTC

- reply

@nprofile…fl4q Sort of, more like ‘config.cache’: https://wiki.samba.org/index.php/Waf#Using_--cross-answers
But I wouldn’t want to manually provide that information.
(I hope BLUE gets that right. :-))

2026-03-21 10:45:15 UTC

When cross-compiling, Waf *requires* QEMU at configure time to do things like computing sizeof(xyz). 🤦
https://wiki.samba.org/index.php/Waf#cross-compiling

2026-03-19 21:42:28 UTC

- reply

@nprofile…dr5p 💯

2026-03-19 08:01:57 UTC

I’m packaging an HPC benchmark and the letters “cray” appear in the hash. Coincidence?
https://toot.aquilenet.fr/system/media_attachments/files/116/254/787/989/279/914/original/285e7ccf6af9c027.png

2026-03-05 17:21:50 UTC

- reply

Inspired by https://codeberg.org/guix/guix/pulls/6097#issuecomment-10343652

2026-03-05 17:21:31 UTC

Send SIGHUG, not SIGKILL.

2026-03-04 17:46:09 UTC

- reply

@nprofile…rwr4 Heh. :-) The thing is that systemd (or OpenRC, or any init system) just sees part of the story; here we get to see all the operating system configuration, whether or not that leads to a service handled by the init system.

2026-03-04 16:04:30 UTC

- reply

@nprofile…nyx8 Tu parles de « comptes administrateurs » au sens de compte root ?

2026-03-03 22:43:46 UTC

- reply

@nprofile…nyx8 Stadire qu’on le voit de très près, le problème.
https://toot.aquilenet.fr/@civodul/115570822148384443

2026-03-03 22:00:53 UTC

L’autre bonne raison de s’opposer à la ZRR intégrale : refuser de participer à la militarisation de la société.
https://zrr.collectif-inria.fr/

2026-03-03 10:31:24 UTC

- reply

@nprofile…wfat It’d really like to see more stuff on this line of work, making it easier to comprehend our systems, so one doesn’t need to have a black belt in Guix and Scheme to understand what’s going on.
That was the whole point of Guix!

2026-03-03 07:42:48 UTC

- reply

@nprofile…h4ph Yes, I had a local copy of the repo. But I think they made it possible to clone previous repositories over SSH.

2026-03-02 22:05:01 UTC

Just salvaged #Guix Explorer from #NotABug, and you know what? It’s still as hypnotic as before. 👇
https://codeberg.org/civodul/guix-explorer

2026-03-01 17:21:02 UTC

📢 Public service announcement: several #Guile repositories formerly at #NotABug (now down) have now moved to #Codeberg 👇
https://codeberg.org/guile-compression
https://codeberg.org/guile-gcrypt
More info: https://lists.gnu.org/archive/html/guile-user/2026-03/msg00000.html

2026-03-01 16:07:00 UTC

- reply

@nprofile…047x In the end it has little to do with code: these things are quickly sabotaging the social fabric before our eyes, be it in free software projects or in other areas.

2026-02-28 18:12:56 UTC

- reply

@nprofile…0f9m Thanks to @nprofile…hdk0, the Guix Data Service provides per-package reproducibility data for each commit, based on info gathered from the two official build farms:
https://data.guix.gnu.org/repository/1/branch/master/latest-processed-revision/package-reproducibility
(This page is currently timing out though.)
Pretty nice because it gives a near real-time view of the situation!

2026-02-26 22:24:08 UTC

- reply

@nprofile…3nd8 L’oligarchie, le complexe militaro-industriel, l’État dans son côté autoritaire et répressif sont en position de force beaucoup plus que dans les années 2000.
Ça fait que les mouvements en ce moment, comme les Soulèvements ou XR ou les collectifs antifa, sont plus dans la réaction que dans la proposition comme au Larzac en 2003.
C’est quand même des belles choses qui demandent qu’on ne se laisse pas aller à la résignation. :-)

2026-02-26 21:43:06 UTC

- reply

@nprofile…56x8 It’s not glibc (it’s a graft, so only glibc itself gets rebuilt): it’s mostly the ‘gnome-team’ and ‘python-team’ branches that had been accumulating world rebuilds that were no longer relevant.
I cleaned that up and we’re back to “only” 60k pending builds. :-)

2026-02-26 21:40:07 UTC

- reply

@nprofile…9muw Wonderful! I’ve dreamed of this for so long!

2026-02-26 21:36:05 UTC

- reply

@nprofile…fl4q Also resonates with me due to my past work in digital typography and authoring, which I really enjoyed, though my contributions (in Lout and Skribilo) were by far not as successful. (For Lout I mimicked the equation formatting algorithm of the TeX Book to improve on what was there but the result was still awful. :-))

2026-02-26 21:32:01 UTC

- reply

@nprofile…fl4q Beautiful.
I also just discovered the 2003 paper you co-authored, and it’s amazing!
https://lilypond.gitlab.io/static-files/media/xivcim.pdf

2026-02-26 20:26:46 UTC

We went from copyright law being shaped by Disney demanding ever-increasing terms, to copyright law being effectively “optional” with OpenAI & co. lobbying so LLM output is not considered a derivative work of its inputs.
A 180° turn.
The law is still shaped by the powerful, but they now have different interests.

2026-02-26 08:28:24 UTC

- reply

@nprofile…l0sz @nprofile…gnen This is one interpretation I’ve read, but I guess it’s a grey area.

2026-02-25 22:04:59 UTC

- reply

@nprofile…wfat Nice job! Different from the prompt/output form we’re used to, but very pleasant. 👍
@nprofile…063y @nprofile…kqpd @nprofile…nx2p @nprofile…7yuj @nprofile…9kfq

2026-02-25 20:30:01 UTC

- reply

@nprofile…gnen The LLM output is public domain. If it’s “legally significant” (10 lines of code or more), and if these LLM-produced contributions are not clearly identified, then one could consider the whole as public domain, AIUI.

2026-02-25 20:26:10 UTC

- reply

@nprofile…lks0 Anthropic claims they spent $20k on the Claude C Compiler (probably underestimated, but that gives an idea).

2026-02-25 18:04:49 UTC

- reply

@nprofile…ygzq This guy called Claude everybody talks about might be willing to help?

2026-02-25 17:59:57 UTC

- reply

@nprofile…047x Because it’s so tempting? And “everyone does it”, and “look how it could benefit our project”.

2026-02-25 17:58:18 UTC

- reply

So yes, maybe we’ll have to give up on some dreams—like the year of the Hurd on the desktop.
But in exchange, we’ll give something more valuable: human beings sharing their passion, helping each other, and building things together. The real asset of free software.

2026-02-25 17:56:24 UTC

- reply

I think we need solidarity. We need to recognize the harms of genAI, including from a free software standpoint.
And we need bigger projects to show the way: to clearly state their rejection, and not on the grounds of quality assurance—a concern bound to become irrelevant—but really on the grounds of ethics, refusing to be part of harm this does to society.

2026-02-25 17:54:16 UTC

- reply

It’s probably easier for a big project like Gentoo to say “no” to genAI—they have enough contributors anyway, they don’t need it.
So what do we do?

2026-02-25 17:53:00 UTC

- reply

I think these two factors—lack of humanpower for a “big” vision—couple with the passion for technicalities typical of such projects make them particularly vulnerable to genAI.
Because yes, “we” want SMP support in Mach and it’s not been happening until this contributor achieved something with the help of genAI.

2026-02-25 17:51:05 UTC

- reply

Like the Emacs-in-Rust thing¹, it’s another pipe dream project with scarce humanpower.
¹ https://toot.aquilenet.fr/@civodul/116099711903317535

2026-02-25 17:49:14 UTC

GenAI reaches another unexpected corner of #FreeSoftware: the #Hurd
https://lists.gnu.org/archive/html/bug-hurd/2026-02/msg00133.html

2026-02-25 17:47:05 UTC

- reply

@nprofile…njfr Yeah, I realized after seeing ‘CLAUDE.md’. It’s really insidious because it’s a regular text file that you would think is meant for humans to read.
@nprofile…8z6r

2026-02-25 15:03:37 UTC

- reply

@nprofile…8z6r D’oh! Maybe! :-)

2026-02-21 17:27:32 UTC

“We estimated 75 200 violent deaths (95% CI 63 600–86 800) between Oct 7, 2023, and Jan 5, 2025, representing approximately 3.4% of the Gaza Strip's pre-conflict population. Women, children (ie, younger than 18 years), and older people (ie, older than 64 years) comprised 56.2% […]. The [Ministry of Health] figure for this period (49 090 violent deaths) was 34.7% below our central estimate.”
https://doi.org/10.1016/S2214-109X(25)00522-4

2026-02-20 21:28:32 UTC

- reply

@nprofile…fl4q @nprofile…063y Agreed! It’s great to have this kind of blog post about the event.

2026-02-20 19:51:13 UTC

This #Guix Ruby extension by @nprofile…uz5f is an inspiring way to provide a nice and smooth developer experience 👇
https://aus.social/@carlozancanaro/116035134690193504

2026-02-20 17:05:08 UTC

- reply

@nprofile…u5g7 A study from the EU Parliament as well as the US Copyright Office claim that code produced with little or no human intervention cannot be copyrighted, making it public-domain.
https://www.europarl.europa.eu/RegData/etudes/STUD/2025/774095/IUST_STU(2025)774095_EN.pdf
https://copyright.gov/ai/Copyright-and-Artificial-Intelligence-Part-2-Copyrightability-Report.pdf

2026-02-19 23:00:25 UTC

- reply

I can relate to the pipe dream kind of free software project—the kind that would require too many person-years but could make such a difference.
But I cannot relate to this process of telling a machine what to produce, crafting checks so it will keep trying until it succeeds. That feels too remote from what I enjoy about hacking and from what I like to share.

2026-02-19 22:53:17 UTC

- reply

Ironic because the process producing this code is itself pretty much harming autonomy: the person behind it gets a lot done, implementing their high-level view, but they sidestep the whole programming experience as we know it—learning and failing, through sweat and tears, but also joy.

2026-02-19 22:47:54 UTC

- reply

I find it ironic given that Emacs was designed with the idea that it would empower people by allow them to learn programming “without noticing”.
https://dspace.mit.edu/bitstream/handle/1721.1/5736/AIM-519A.pdf
https://toot.aquilenet.fr/system/media_attachments/files/116/099/728/262/651/266/original/9bd262382cc20b2a.png

2026-02-19 22:43:34 UTC

Here we are: an Emacs clone (a rewrite in Rust!) being vibe-coded
https://github.com/eval-exec/neomacs/
🧵

2026-02-19 14:02:54 UTC

Sarah Cohen-Boulakia : « On publie trop et trop vite »
https://themeta.news/sarah-cohen-boulakia/
#RechercheReproductible

2026-02-19 13:19:10 UTC

- reply

Le 19 mai sera consacrée à #Guix comme outil pour la #RechercheReproductible avec des retours d’expérience et des ateliers pour comprendre les enjeux et pour débuter :
https://jrfrr-2026.sciencesconf.org/resource/page/id/2

2026-02-19 13:15:50 UTC

Les Journées du Réseau de recherche reproductible @nprofile…rw7a auront lieu du 19 au 21 mai à Bordeaux 👇
https://jrfrr-2026.sciencesconf.org/
C’est le moment de s’inscrire !
#RechercheReproductible #ScienceOuverte #ESR

2026-02-09 14:15:06 UTC

Nice poster on using #Guix for #ReproducibleResearch 👇
https://zenodo.org/records/17435294/files/Long_lasting_bit-reproducible_research_with_GNU_Guix.pdf

2026-02-09 14:00:13 UTC

- reply

@nprofile…yfhg Yes that proposal looked really interesting! It’s a bit over my head so I didn’t offer to sponsor it, but it would be great to revive it or use it as a starting point.

2026-02-08 22:55:24 UTC

Je recommande le film « À Pied d’œuvre », adaptation du bouquin que j’avais déjà bien aimé de Franck Courtès (aucun lien de parenté mais je suis solidaire :-)).

2026-02-08 22:42:47 UTC

- reply

@nprofile…56x8 Downside is that you cannot boot old generations, can you?

2026-02-08 10:25:45 UTC

- reply

@nprofile…063y What⁈ What happened with justice.gov?

2026-02-07 20:55:52 UTC

- reply

@nprofile…047x Yeah. With these kinds of announcement, they make it sound like human creativity is worthless, but it’s this very creativity and the commons that were built that made those machines possible.

2026-02-07 20:53:41 UTC

Funny that GDB and Bash bundle the documentation of Readline (rluser.texi and hsuser.texi) for inclusion in their own manual.

2026-02-07 10:56:34 UTC

Destruction is what these companies specialize in.
https://cosocial.ca/@mhoye/116024404796795456

2026-02-06 23:34:26 UTC

- reply

@nprofile…ejm2 Source: https://www.ohchr.org/en/press-releases/2026/02/germany-un-expert-warns-space-freedom-expression-shrinking-amidst-growing

2026-02-04 09:16:30 UTC

- reply

Many thanks to @nprofile…affn and all the devroom volunteers for the great experience! 🙏

2026-02-04 09:13:17 UTC

📺 My talk in the #HPC track at #FOSDEM, “Package management in the hands of users: dream and reality” 👇
https://fosdem.org/2026/schedule/event/QKNMJN-package-management-in-the-hands-of-users/
#Guix

2026-02-04 00:01:49 UTC

- reply

@nprofile…ygzq Reverse job interview is when the person you’re employing conducts an interview of yourself. (I was off guard.)

2026-02-01 14:49:07 UTC

Very energetic and compelling talk by Helena Vela Beltran on #EESSI in the #HPC devroom. 👍

2026-02-01 14:26:33 UTC

Great talk by Harmen Stoppels on #Spack in the #HPC devroom at #FOSDEM, starting by stating how Spack deals with the platonic ideal of packages. :-)

2026-02-01 09:52:58 UTC

- reply

Demo 👉 https://people.bordeaux.inria.fr/lcourtes/tmp/doc/

2026-02-01 09:51:48 UTC

- reply

@nprofile…kqpd Regarding the online copy of manuals, check this out 👇
https://codeberg.org/guix/maintenance/pulls/69
It won’t be as good as Racket’s (yet), but at least it brings syntax highlighting and hyperlinks on identifiers in Scheme snippet, and cross-manual links that are not 404. :-)

2026-02-01 09:44:05 UTC

- reply

@nprofile…kqpd Wait wait wait, not only did Jessica implement a VSCode plugin to properly indent Scheme, we also get a live demo of it!

2026-02-01 09:32:37 UTC

And how @nprofile…kqpd touching on #Guile development outside of #Emacs—which is kinda crucial: #Guix, #Guile, #Spritely folks certainly want to reach out to people not into it!

2026-02-01 09:26:16 UTC

- reply

I think @nprofile…8z6r also got all the #Guile hackers hooked by showing the beautiful backtraces that BLUE displays, a long way from what we know al too well.

2026-02-01 09:21:17 UTC

- reply

The room is also full of parens!

2026-02-01 09:19:06 UTC

Minimalistic Computing devroom is now packed, and there’s already been much #Guile goodness: @nprofile…8z6r showcased the amazing BLUE build system, and @nprofile…wfat is wrapping up on REPL-driven development with Ares/Arei!
#FOSDEM