Stories
Slash Boxes
Comments

SoylentNews is people

SoylentNews is powered by your submissions, so send in your scoop. Only 13 submissions in the queue.
posted by mrcoolbp on Monday October 05, @09:25AM   Printer-friendly
from the corporate-Codes-of-Censorship dept.

The editor-in-chief at the Systemd-free Linux Community has reposted a thorough analysis of the current situation with most GNU/LInux systemd/Linux forums and projects. Nowadays it is basically impossible to find a proper distro which retains a healthy community and/or is able to avoid systemd, elogind, and the other corporate problem children.

One commenter’s traumatic life with linux (editor in chέf: best article posted here ever )

I wished you were not anonymous and used a nick-name somewhere so we can at least refer to your wonderful writing without ambiguity.

Comment left on the short list of distros without elogind

Long time reader first time commenter to your blog. I wanted to add my two cents as a looooong time user of Linux. First with Mandrake Linux all the way back in the mid-late 90s then later on Gentoo. For over 20 years I maintained my own home build server and ran Gentoo on all my desktops, home servers and web servers. Well I did have some servers and devices running a mix of Free/Open/NetBSD as well to of course depending on what I was using them for.

[...] Even within the OpenBSD project lately I'm finding a lot of flak. The actual code, developers and mailing lists are still good. But pretty much everything around it like the IRC channels are getting flooded with people that do the things I've mentioned above. They'll sit in there all day pushing an off-topic agenda but the moment you reply to them or call them out they flex their moderation powers to ban you for being "off topic". Same goes for their sub-reddit. I don't think Theo will allow something like a CoC any time soon but once he's gone I do not think the project will be able to survive for very long. It's also getting _much_ harder to port over software due to all the garbage happening in the Linux/FreeBSD projects. On top of that many people that obviously do not speak English as a first language are showing up as of late and trying to submit patches/drivers/ports generated using LLM software. In other words: garbage untested code. Right now none of that is allowed and AI code within the project is banned since it's impossible to know what code was used to train said AI. But such people have already caused massive problems for the public CVS servers by flooding them with scraping bot requests. Then on top of that there has been a flood of idiots on youtube lately promoting OpenBSD as the "true hacker OS". You know the kind. The people that post a screenshot of their desktop+neofetch and try to game the youtube/google adsense system for money. They've all moved on from promoting Arch/Gentoo Linux to promoting things like OpenBSD lately. They're drawing in a ton of idiots in their wake. It's obvious none of them actually use the software.

systemd was contentious from the beginning, having entered the Linux communities at large via Debian despite not being in the top three choices when Debian TC voting occurred.

Previously:
(2026) Age Checks Creep Into Linux, systemd Locks It in, Developer Defends Himself
(2026) Linux From Scratch Abandoning SysVinit Support
(2025) Devuan Version 6.0 "Excalibur" is Available Now
(2022) Systemd Creator Lands At Microsoft
(2020) Modern Versions of systemd Can Cause an Unmount Storm During Shutdowns
(2020) At Last, the Fix No One Asked for: Portable Home Directories Merged into systemd
(2019) Debian Packagers Voted "Systemd but We Support Exploring Alternatives"
(2017) Savaged by Systemd
(2017) Linux systemd Gives Root Privileges to Invalid Usernames
(2017) Review of Devuan 1.0.0
(2016) SystemD Mounts EFI pseudo-fs RW, Facilitates Permanently Bricking Laptops, Closes Bug Invalid
(2015) A Technical Critique of Systemd
(2014) Linus Torvalds Shares his Opinion on systemd
... and more.


Original Submission

Related Stories

Linus Torvalds Shares his Opinion on systemd 57 comments

Systemd has turned into the Godzilla of Linux controversies. "Everywhere you look it's stomping through blogs, rampaging through online discussion threads, and causing white-hot flames that resemble Godzilla's own breath of death," writes Jim Lynch. Now Sam Varghese reports at iTWire that although Linus Torvalds is well-known for his strong opinions, when it comes to systemd, Torvalds is neutral. "When it comes to systemd, you may expect me to have lots of colorful opinions, and I just don't," says Torvalds. "I don't personally mind systemd, and in fact my main desktop and laptop both run it."

Oh, there's been bitter fights before. Just think about the emacs vs vi wars. Or, closer to systemd, the whole "SysV init" vs "BSD init" differences certainly ended up being things that people had "heated discussions" about. Or think about the desktop comparisons.

I'm not really sure how different the systemd brawls are from those. It's technical, but admittedly the systemd developers have also been really good at alienating people on a purely personal level too. Not that that is anything particularly new under the sun _either_: the (very) bitter wars between the GPL and the BSD license camps during late-80s and early-90s were almost certainly more about the persons involved and how they pissed off people than necessarily deeply about other differences (which existed, obviously, but still).

Torvalds was asked if systemd didn't create a single point of failure which makes a system unbootable if it fails. "I think people are digging for excuses. I mean, if that is a reason to not use a piece of software, then you shouldn't use the kernel either."

A Technical Critique of Systemd 73 comments

Structural and semantic deficiencies in the systemd architecture for real-world service management

This is a in-depth architectural critique of systemd. It claims to be the first purely technical review of systemd internals, and provides a detailed analysis of several components. It criticizes on the basis of ordering related failures, a difficult to predict execution model, non-determinism in boot-order, as well as several other points.

Though many users would perceive the long processing pipeline to increase reliability and be more "correct" than the simpler case, there is little to acknowledge this. For one thing, none of jobs, transactions, unit semantics or systemd-style dependencies map to the Unix process model, but rather are necessary complications to address issues in systemd being structured as an encapsulating object system for resources and processes (as opposed to a more well-defined process supervisor) and one accommodating for massive parallelism. Reliability gains would be difficult to measure, and that more primal toolkits like those of the daemontools family have been used in large-scale deployments for years would serve as a counterexample needing overview.


Original Submission #1  Original Submission #2 

SystemD Mounts EFI pseudo-fs RW, Facilitates Permanently Bricking Laptops, Closes Bug Invalid 145 comments

A number of users have reported that running "rm --no-preserve-root -rf /" not only deletes all their files (as expected), but also permanently bricks their computers (which is not). Tracing the issue revealed that the ultimate cause was that SystemD mounted the EFI pseudo-fs as read-write even when this FS was not listed in fstab, and deleting certain files in this pseudo-fs causes certain buggy, but very common, firmware not to POST anymore. A user reported this bug on SystemD's GitHub issue tracker, asking that the FS be mounted read-only instead of read-write, and said bug was immediately closed as invalid. The comment thread for the bug was locked shortly after. Discuss.

Links:
https://github.com/systemd/systemd/issues/2402
http://thenextweb.com/insider/2016/02/01/running-a-single-delete-command-can-permanently-brick-laptops-from-inside-linux/


Original Submission

Review of Devuan 1.0.0 60 comments

Jesse Smith reports via DistroWatch

Conclusions

On the whole, the Devuan project appears to have achieved its goals. The distribution offers users an operating system virtually identical to Debian 8, but with systemd replaced with SysV init. The project provides existing Debian users a clean and easy migration path to Devuan that has only a minimal amount of side effects. Taken on its own, Devuan is a lightweight operating system with a fairly minimal (and responsive) desktop environment.

While Devuan has reached its goals, I had two significant concerns about the distribution. The first concern was the system installer. While it worked, I'm curious as to why Devuan appears to have discarded the reliable Debian installer in favour of a less feature rich and less polished installation process. Other Debian-friendly installers, such as the one which ships with Linux Mint Debian Edition, are available if a more streamlined approach is wanted.

My other concern is that Devuan 1.0.0 is about two years behind Debian. A fork of Debian without systemd seemed promising and interesting in 2015 when Debian 8 was released. But now, two years later, with Debian 9 on the horizon, Devuan 1 feels outdated. The software, such as the office suite and kernel, are about three years old at this point and unlikely to appeal to any except the most conservative users. The distribution may hold more appeal on servers where change often happens more slowly, but even there some of the Devuan packages are starting to show their age.

At this point I suspect Devuan 1 will only appeal to the more enthusiastic members of the anti-systemd crowd. If Devuan 2 can be launched shortly after Debian 9 comes out later this year then I could see the project gaining a stronger user base, but at the moment Devuan feels like an interesting idea that took too long to get off the ground.

Previous: Devuan Stable Release -- at Last!

[Editors Note: Debian 9 has been released. We ran a story on it a few hours ago.]


Original Submission

Linux systemd Gives Root Privileges to Invalid Usernames 130 comments

Submitted via IRC for TheMightyBuzzard

A bug in Linux's systemd init system causes root permissions to be given to services associated with invalid usernames, and while this could pose a security risk, exploitation is not an easy task.

A developer who uses the online moniker "mapleray" last week discovered a problem related to systemd unit files, the configuration files used to describe resources and their behavior. Mapleray noticed that a systemd unit file containing an invalid username – one that starts with a digit (e.g. "0day") – will initiate the targeted process with root privileges instead of regular user privileges.

Systemd is designed not to allow usernames that start with a numeric character, but Red Hat, CentOS and other Linux distributions do allow such usernames.

"It's systemd's parsing of the User= parameter that determines the naming doesn't follow a set of conventions, and decides to fall back to its default value, root," explained developer Mattias Geniar.

While this sounds like it could be leveraged to obtain root privileges on any Linux installation using systemd, exploiting the bug in an attack is not an easy task. Geniar pointed out that the attacker needs root privileges in the first place to edit the systemd unit file and use it.

[...] Systemd developers have classified this issue as "not-a-bug" and they apparently don't plan on fixing it. Linux users are divided on the matter – some believe this is a vulnerability that could pose a serious security risk, while others agree that a fix is not necessary.

See, this is why we can't have nice init systems.

Source: http://www.securityweek.com/linux-systemd-gives-root-privileges-invalid-usernames


Original Submission

Savaged by Systemd 31 comments

The other day, Michael W Lucas, who is normally known for good technical literature, put up a wild experiment of a short story: Savaged by Systemd. It's erotica, sort of. It's computer erotica, to be specific. It's Linux sysadmin erotica, to be more specific. OK, fine, it's systemd erotica. Really. Anyway, despite the subject and the genre, and in spite of the combination of the two, the e-book is trending and rising in quite a few lists.

Hopefully he can still remain focused on Absolute FreeBSD and be able to get that finished by the next BSDCan.

[Ed note: Has anybody actually bought and read this short story? I wasn't going to spend $2.99 to see what the hubub was about. - cmn32480]


Original Submission

Debian Packagers Voted "Systemd but We Support Exploring Alternatives" 81 comments

Earlier this month we covered a story where we quoted the following:

Due to Debian developers having differing opinions on handling non-systemd bugs in 2019 and the interest/commitment to supporting systemd alternatives in the scope of Debian packaging and various related friction points, they've taken to a new general resolution over weighing init system diversity.

The results of the ballot to discover how the Debian devs feel about their earlier decision with the benefit of hindsight is now in:

Using its power under Constitution section 4.1 (5), the project issues the following statement describing our current position on Init systems, multiple init systems, and the use of systemd facilities. This statement describes the position of the project at the time it is adopted. That position may evolve as time passes without the need to resort to future general resolutions. The GR process remains available if the project needs a decision and cannot come to a consensus.

The Debian project recognizes that systemd service units are the preferred configuration for describing how to start a daemon/service. However, Debian remains an environment where developers and users can explore and develop alternate init systems and alternatives to systemd features. Those interested in exploring such alternatives need to provide the necessary development and packaging resources to do that work. Technologies such as elogind that facilitate exploring alternatives while running software that depends on some systemd interfaces remain important to Debian. It is important that the project support the efforts of developers working on such technologies where there is overlap between these technologies and the rest of the project, for example by reviewing patches and participating in discussions in a timely manner.

Packages should include service units or init scripts to start daemons and services. Packages may use any systemd facility at the package maintainer's discretion, provided that this is consistent with other Policy requirements and the normal expectation that packages shouldn't depend on experimental or unsupported (in Debian) features of other packages. Packages may include support for alternate init systems besides systemd and may include alternatives for any systemd-specific interfaces they use. Maintainers use their normal procedures for deciding which patches to include.

Debian is committed to working with derivatives that make different choices about init systems. As with all our interactions with downstreams, the relevant maintainers will work with the downstreams to figure out which changes it makes sense to fold into Debian and which changes remain purely in the derivative.


Original Submission

At Last, the Fix No One Asked for: Portable Home Directories Merged into systemd 110 comments

The systemd-homed service, which enables portable home directories, has been merged into the code for systemd and will be included in the forthcoming 245 release.

Systemd releases are typically every three to four months, and version 244 was finalised at the end of November 2019. The new merge includes over 21,000 additions to systemd. Once 245 is out, it will be up to individual Linux distributions to decide when to update it. Use of the new home directories service is optional.

[...]Each directory it manages encapsulates both the data store and the user record of the user so that it comprehensively describes the user account, and is thus naturally portable between systems without any further, external metadata.

Home directories in the new system support several storage mechanisms and may be located on a removeable drive. The user record is cryptographically signed so the user cannot modify it themselves without invalidating it. There is an option for encryption with fscrypt (applies encryption at the directory level), or mounting from a CIFS network share, or in a partition encrypted with LUKS2 (Linux Unified Key Setup). This last is the most secure approach.

[...]One use case is where a user has a PC running Linux in both their home and office, and is able to carry their home directory with them on a portable storage device. The advent of cloud storage has made this less of a problem than would have been the case a few years back, and a common reaction to the new systemd approach is that the problems it fixes are not pressing and may be outweighed by potential incompatibilities. ®

See the file: https://github.com/poettering/systemd/blob/homed/docs/HOME_DIRECTORY.md for more information on what systemd-homed is intended to do, an how.


Original Submission

Modern Versions of systemd Can Cause an Unmount Storm During Shutdowns 102 comments

System adminsitrator Chris Siebenmann has found Modern versions of systemd can cause an unmount storm during shutdowns:

One of my discoveries about Ubuntu 20.04 is that my test machine can trigger the kernel's out of memory killing during shutdown. My test virtual machine has 4 GB of RAM and 1 GB of swap, but it also has 347 NFS[*] mounts, and after some investigation, what appears to be happening is that in the 20.04 version of systemd (systemd 245 plus whatever changes Ubuntu has made), systemd now seems to try to run umount for all of those filesystems all at once (which also starts a umount.nfs process for each one). On 20.04, this is apparently enough to OOM[**] my test machine.

[...] Unfortunately, so far I haven't found a way to control this in systemd. There appears to be no way to set limits on how many unmounts systemd will try to do at once (or in general how many units it will try to stop at once, even if that requires running programs). Nor can we readily modify the mount units, because all of our NFS mounts are done through shell scripts by directly calling mount; they don't exist in /etc/fstab or as actual .mount units.

[*] NFS: Network File System
[**] OOM Out of memory.

We've been here before and there is certainly more where that came from.

Previously:
(2020) Linux Home Directory Management is About to Undergo Major Change
(2019) System Down: A systemd-journald Exploit
(2017) Savaged by Systemd
(2017) Linux systemd Gives Root Privileges to Invalid Usernames
(2016) Systemd Crashing Bug
(2015) tmux Coders Asked to Add Special Code for systemd
(2016) SystemD Mounts EFI pseudo-fs RW, Facilitates Permanently Bricking Laptops, Closes Bug Invalid
(2015) A Technical Critique of Systemd
(2014) Devuan Developers Can Be Reached Via vua@debianfork.org
(2014) Systemd-resolved Subject to Cache Poisoning


Original Submission

Systemd Creator Lands At Microsoft 65 comments

Good News Everyone!

Systemd Creator Lands At Microsoft

Yesterday's surprise was that Lennart Poettering quietly had left Red Hat following a decade and a half there leading PulseAudio among other projects and ultimately going on to start systemd that has fundamentally reshaped modern Linux distributions. It turns out he had joined Microsoft and continuing his work on systemd.

After yesterday's article about Lennart no longer being at Red Hat, I began receiving tips that the systemd creator had some time back quietly joined Microsoft along with various public comments on Twitter and other mediums by individuals suggesting he joined the Redmond company... At first I thought they were jokes or just snarky remarks, but after a day of following up with folks, it actually turns out not to be a joke.

The prominent open-source developer responsible for several prominent projects joined Microsoft and continuing his focus on systemd development.

[Ed's Comment: AC Friendly withdrawn. You can blame you-know-who for the spamming]

Devuan Version 6.0 "Excalibur" is Available Now 10 comments

https://itsfoss.com/news/devuan-6-release/

Devuan is a Linux distribution that takes a different approach from most popular distros in the market. It is based on Debian but offers users complete freedom from systemd.

The project emerged in 2014 when a group of developers decided to offer init freedom. Devuan maintains compatibility with Debian packages while providing alternative init systems like SysVinit and OpenRC.

With a recent announcement, a new Devuan release has arrived with some important quality of life upgrades.

Codenamed "Excalibur", this release arives after extensive testing by the Devuan community. It is based on Debian 13 "Trixie" and inherits most of its improvements and package upgrades.

Devuan 6.0 ships with Linux kernel 6.12, an LTS kernel that brings real-time PREEMPT_RT support for time-critical applications and improved hardware compatibility.

On the desktop environment side of things, Xfce 4.20 is offered as the default one for the live desktop image, with additional options like KDE Plasma, MATE, Cinnamon, LXQt, and LXDE.

The package management system gets a major upgrade with APT 3.0 and its new Solver3 dependency resolver. This backtracking algorithm handles complex package installations more efficiently than previous versions. Combined with the color-coded output, the package management experience is more intuitive now.

This Devuan release also makes the merged-/usr filesystem layout compulsory for all installations. Users upgrading from Daedalus (Devuan 5.0) must install the usrmerge package before attempting the upgrade.

Similarly, new installations now use tmpfs for the /tmp directory, storing temporary files in RAM instead of on disk. This improves performance through faster read and write operations.

And, following Debian's lead, Devuan 6.0 does not include an i386 installer ISO. The shift away from 32-bit support is now pretty much standard across major distributions. That said, i386 packages are still available in the repositories.

The next release, Devuan 7, is codenamed "Freia". Repositories are already available for those adventurous enough to be early testers.

This release supports multiple CPU architectures, including amd64, arm64, armhf, armel, and ppc64el. You will find the relevant installation media on the official website, which lists HTTP mirrors and torrents.

Existing Devuan 5.0 "Daedalus" users can follow the official upgrade guide.


Original Submission

Linux From Scratch Abandoning SysVinit Support 33 comments

Linux From Scratch Abandoning SysVinit Support:

Linux From Scratch was one of the holdouts continuing optional SysVinit init system support through 2026, but that's now ending. Linux From Scratch "LFS" and Beyond Linux From Scratch "BLFS" are ending their System V Init support moving forward.

LFS/BLFS will not be supporting System V in future versions due to the increasing demands of continuing to maintain and test both System V and systemd support. LFS/BLFS is a volunteer effort and continuing to maintain the SysVInit support is a burden. Additionally, with GNOME and soon KDE Plasma going to depend upon functionality explicitly found in systemd and not System V, it's going to be all the more challenging.

Bruce Dubbs of Linux From Scratch announced the decision:

"There are two reasons for this decision. The first reason is workload. No one working on LFS is paid. We rely completely on volunteers. In LFS there are 88 packages. In BLFS there are over 1000. The volume of changes from upstream is overwhelming the editors. In this release cycle that started on the 1st of September until now, there have been 70 commits to LFS and 1155 commits to BLFS (and counting). When making package updates, many packages need to be checked for both System V and systemd. When preparing for release, all packages need to be checked for each init system.

The second reason for dropping System V is that packages like GNOME and soon KDE's Plasma are building in requirements that require capabilities in systemd that are not in System V. This could potentially be worked around with another init system like OpenRC, but beyond the transition process it still does not address the ongoing workload problem.

He ended the announcement with:

"As a personal note, I do not like this decision. To me LFS is about learning how a system works. Understanding the boot process is a big part of that. systemd is about 1678 "C" files plus many data files. System V is "22" C files plus about 50 short bash scripts and data files. Yes, systemd provides a lot of capabilities, but we will be losing some things I consider important.

However, the decision needs to be made."


Original Submission

Age Checks Creep Into Linux, systemd Locks It in, Developer Defends Himself 21 comments

https://www.theregister.com/2026/03/24/foss_age_verification/?td=rt-3a

From TFA:

After weeks of debate, code to record user age was finally merged into the Linux world's favorite system management daemon.

Pull request #40954 to the systemd project is titled "userdb: add birthDate field to JSON user records." It's a new function for the existing userdb service, which adds a field to hold the user's date of birth.

[...] The change comes after the recent release of systemd 260 but unless it is reverted for some reason, it will be part of systemd 261. One of the justifications is to facilitate the new parental controls in Flatpak, which are still in the draft stage.

[...] The TBOTE findings suggest that Meta is the biggest donor behind the lobbying for these age-verification laws and the App Store Accountability Act (ACCA). TBOTE claims it has directly traced more than $25 million, and that Meta could have spent upward of $2 billion on this over the last year. It also points to €10 million-plus spent lobbying in Europe.

In the US, the main group pushing for these laws is the relatively young Digital Childhood Alliance (DCA). As right-wing think tank the "Institute for Family Studies" reported a year ago, this was assembled by over 50 conservative groups. Six months later, in July 2025, Bloomberg also reported that Meta was funding the DCA. For such a young and small organization, the DCA certainly seems to have had a rapid and almost disproportionate impact.

Nuff said.

The Engineer Who Tried to Put Age Verification Into Linux

In March of 2026, systemd, the init system that boots most modern Linux distributions, merged a pull request adding a birthDate field to its user database. The stated purpose was compliance with California's AB-1043, Colorado's SB26-051, and Brazil's Lei 15.211/2025, a wave of age verification laws requiring operating systems to collect birth dates from users at account setup, then feed that data to app stores via a real-time API. The PR was submitted by a contributor using the GitHub handle dylanmtaylor. Within days it had 945 comments and was locked by maintainers. Someone opened a revert PR. Lennart Poettering closed it without merging on March 19th, saying the field is optional and systemd "enforces zero policy." The birthDate field is still in systemd. systemd PR #40954 revert PR #41179

The lasting damage was knowing it could happen at all: that a single contributor with no stated organizational backing could submit compliance infrastructure for surveillance law directly into the software that boots your computer, get it merged by two Microsoft employees, and have the creator of systemd personally block the removal.

This discussion was created by mrcoolbp (68) for logged-in users only, but now has been archived. No new comments can be posted.
Display Options Threshold/Breakthrough Mark All as Read Mark All as Unread
The Fine Print: The following comments are owned by whoever posted them. We are not responsible for them in any way.
(1)
  • (Score: 4, Insightful) by turgid on Monday October 05, @09:42AM (14 children)

    by turgid (4318) Subscriber Badge on Monday October 05, @09:42AM (#1455700) Journal

    Go back in the git history to before LLMs were invented. Fork. Leave the loonies to their madhouses.

    • (Score: 5, Funny) by JoeMerchant on Monday October 05, @11:41AM (12 children)

      by JoeMerchant (3937) on Monday October 05, @11:41AM (#1455708) Journal

      Yeah, this gunpowder stuff is madness - it blows up in your face all the time. Let's retreat to our stone walled fortresses and defend them properly with bows, arrows and spears.

      You don't hear much about that, because of who writes history.

      --
      🌻🌻🌻🌻✌️ [google.com]
      • (Score: 5, Interesting) by turgid on Monday October 05, @11:46AM (10 children)

        by turgid (4318) Subscriber Badge on Monday October 05, @11:46AM (#1455709) Journal

        Remember what Linus said about letting C++ into the kernel? The more I see of this vibe coded stuff the more I see the people who do it as the new C++ programmers. Design? Nah. Maintainability? Who needs that, just feed it back into the AI! Testing? A machine wrote it, it must be right but here are some superficial unit tests to placate you. Bloat? Just buy a bigger computer. Language? Everything's Python these days anyway...

        I jest, but only partially. Try not to take it personally.

        • (Score: 4, Touché) by Anonymous Coward on Monday October 05, @12:17PM (4 children)

          by Anonymous Coward on Monday October 05, @12:17PM (#1455713)

          I realize you hate LLMs, but look at the timeline. All of this crap came about years before AI/LLMs.

          • (Score: 2) by turgid on Monday October 05, @12:23PM (1 child)

            by turgid (4318) Subscriber Badge on Monday October 05, @12:23PM (#1455714) Journal

            The stupidity problems can be solved by forking the projects, whether the cause is human PHB-sim, bandwagon riding or indeed LLMs. When I am rich enough not to have to work anymore, I hope to be able to devote some time to fixing some of the worst stuff. So don't hold your breath.

            • (Score: 0) by Anonymous Coward on Monday October 05, @01:28PM

              by Anonymous Coward on Monday October 05, @01:28PM (#1455721)

              It's still humans engaging / employing LLMs. I agree with forking, but you lessen and maybe dilute your worker pool, which might be a good thing.

          • (Score: 2) by turgid on Monday October 05, @12:24PM (1 child)

            by turgid (4318) Subscriber Badge on Monday October 05, @12:24PM (#1455715) Journal

            By the way, I'm good friends with Claude these days. I don't hate him/her/it.

            • (Score: 4, Interesting) by JoeMerchant on Monday October 05, @12:40PM

              by JoeMerchant (3937) on Monday October 05, @12:40PM (#1455717) Journal

              Claude has improved as "a friend" a lot in the last two years.

              Wild tangent: if "Poor Things" is still streaming on anything you have access to, it's a fun story which has a core element of "growing up fast" ... and a lot of other things, read the synopsis before starting, I don't want to be a spoiler but you should have _some_ idea of what you're getting into before starting that one.

              I perceived potential in Claude back in 2025, and I have not been disappointed with the progress.

              --
              🌻🌻🌻🌻✌️ [google.com]
        • (Score: 2) by JoeMerchant on Monday October 05, @12:30PM (3 children)

          by JoeMerchant (3937) on Monday October 05, @12:30PM (#1455716) Journal

          Progress isn't a linear thing.

          Bloat is a result of laziness, and there's a lot of that out there - laziness is also a virtue in an industry with far more work than workers, which we've had for 50+ years. Ideas are easy, commercially implemented product is expensive and often takes years to build properly, if you can be lazy about something and move on to the next, that gets things done faster (until interest on your tech debt overwhelms the advantages of laziness...)

          Personally, I have more than a few extreme examples of laziness in my home / hobby projects - and I find that those lazily implemented artifacts are a good basis for LLMs to address the tech debt in: clean up and flesh out the documentation, write proper requirements and specs from, then re-architect and cleanup the implementation. An extreme example is a little weather display that I implemented with the display as a client for a remote server - it was a lazy evolution and it worked well for 3+ years, but anytime there was a power outage I had to re-wake the client because it would get hung when the server wasn't present and the server was much slower to restart. Six or seven restarts was a lot cheaper than "fixing it" - but, then, the external weather server I was using went away and I had to re-code to a new API if I wanted my widget to live... by this time LLM agents were getting pretty good, so I told it to "fix it" - and gave it pointers along the way, and in 3-4 hours of wall clock time (less than an hour of my actual attention), the whole thing was re-implemented - I don't even remember which language it's in now, probably python, doesn't matter to me- I just want my weather widget on the wall updating once every 5 minutes and it does that, direct from the Pi Pico with no intermediate server in the house. It took care of importing pretty fonts into the Pico, optimizing the memory usage so the whole thing fits within the limited resources of the Pico, and now when it reboots it connects directly to the outside weather server once per 5 minutes, with robust retries that don't get hung up if, for instance, the router is slow to assign it an IP address while it sits there booted and ready long before anything else on the network has rebooted.

          So, bloat happens. If it's a problem, it can be addressed, but addressing bloat costs the same thing that causes bloat in the first place: competion for the time and resources required to reduce the bloat. If people turn their LLM tools toward reducing bloat, they can help with that like they can help with creating new (bloated) stuff. It's a matter of priorities, and so far the LLMs seem to be following instructions about what to focus on.

          --
          🌻🌻🌻🌻✌️ [google.com]
          • (Score: 3, Interesting) by turgid on Monday October 05, @01:03PM (1 child)

            by turgid (4318) Subscriber Badge on Monday October 05, @01:03PM (#1455718) Journal

            What's important to me, personally as a human being, is having some sort of understanding of what's going on inside. Professionally, the stuff I work on has to be fairly deterministic and efficient. I need to be able to understand it fairly intimately. Many people do not have these requirements but also they don't understand that people like me exist and that projects like mine exist. They are intent on convincing me that I should give up and just use LLMs. I can assure them that LLMs will not be fit for purpose for a couple more years. At which point, I'll cash out my shares and retire. I am not wasting my personal time on a wild goose chase. I have eager young colleagues to do that. I'll focus on getting the job done and getting the hell out of the crazy work market. I can't allow myself to be distracted by fads.

            • (Score: 3, Interesting) by JoeMerchant on Monday October 05, @02:03PM

              by JoeMerchant (3937) on Monday October 05, @02:03PM (#1455725) Journal

              >What's important to me, personally as a human being, is having some sort of understanding of what's going on inside.

              Back in Atari 800 / C64 / Fruity-2 days, I was studying electrical/computer engineering, and building understanding of the fundamentals... while I worked with BJT response curves and have a crude understanding of how to transform those into NAND and other gates, I never really developed interest into the silicon doping and other arts which make construction of the transistors from raw material (and etching on wafers) possible. I know that wafers can be built and etched to remove layers to form gates and connections, but again - the mechanics of that are extending just a little below what I, personally, found interesting so I started "glossing" at that level. I also am not too jazzed about working in and around clean rooms all the time so... that seemed to be good knowledge to avoid becoming unusually well versed in before getting a job...

              Now: cross-coupling nand gates to make flip-flop memory, that's pretty cool stuff. The different ways to clock and clear and make shift-in state read/write of complex digital thingies, I liked that well enough to get a graduate TA position teaching the digital design lab. I got a much better handle on setup and hold times and related concerns as a result of explaining it to five semesters of students. Earlier in my education I thought I had a pretty good grasp of "how it all worked" in CPUs for the registers, compare circuits, adders, multipliers, etc. but the last piece of the puzzle that I remained mystified about was: memory. How in the hell can a CPU "address" any one of 65536 possible bytes and get that data loaded into a working register? The short answer eventually revealed to me: multiplexors - big nested 16 level multiplexors. But... that's a damn lot of multiplexors... yeah, it is, and that's one of the reasons that memory is as expensive as it is, because it needs that multiplexor addressing circuitry in order to address the storage.

              So, yeah, from that level you can microcode and build instruction sets and assembly langauge and higher level languages on that. A couple of years after graduation, the higher level compilers finally advanced to a point where coding assembly by hand was, like transistor doping chemistry, no longer interesting to me in terms of building systems. It's reassuring that I still hazily remember all those layers down, and I suppose if I find myself immortal in a post-technological reboot of society I might eventually build up digital computing from raw materials again - but that seems to be a rather unlikely future. In the past decade or so, I have been focused more on meta-system definition in data, first XML files, later JSON - leveraging communication meshes on top of MQTT and/or AMQP message brokers, and now movign into direct peer to peer messaging with the system definition of both persistent property values and ephemeral event messages communicated both best-effort and guaranteed delivery, as needed, among various nodes of the larger system. I built my own code generator in C++ which interprets the XML into C / C++ code for the various nodes so you just change a definition in the XML and the C / C++ code for all affected nodes in the system is regnerated to implement the change you specified in the XML... I don't think about the underlying cross-coupled nand gates anymore during "the work," but they're still cool on their own as concepts.

              And, then there's the reality of mentoring where you get on a conference call with screen-share and point out to the baffled junior engineer (with a Senior in his title, rewarding 2 years of presence on the job) that the password isn't working because the 0 should be a O, but IT has locked us out of editing files on USB sticks with our corporate issued laptops, so we need to end run them by opening the maintenance interface of our product which does permit normal USB access...

              LLMs don't solve all the problems, or do all the work, but they have built me a much more capable .mp3 (and .ogg and other formats) music player than I built for myself. A year ago they were too frustrating for me, but in the interim they have improved a lot, and candidly I have also learned how to use them better too. Even moving slowly and carefully and addressing tech debt as I go, the LLM assisted development is moving faster on wall clock time than me moving faster, sloppier, and more lazily (unit tests? I don't need no steenking unit tests.) without them. Then there's the reduced attention / focus demand - an LLM is running in the background as I type this, making progress without as much demand on my mental focus as me coding the solution requires. It's not exactly the "we're all managers now" world that some of the press is making it out to be, but it's in that direction.

              If you're ready for retirement and feel about LLMs the way I feel about BJT doping chemistry, that's totally understandable and I'm sure you can contribute value for years to come without using LLMs - much as office workers continued to contribute value without using PCs through the 80s and early 90s... for me, LLMs enable me to scratch itches that I couldn't reach without them, and maybe continue to do "techie stuff" a few more years into my twilight when life demands and ADHD no longer allow me to "get in the zone" I used to be in 30+ hours a week for 2-4 hours at a stretch, today I'm lucky to get 15 minutes at a stretch before some demand or another interrupts any focused activity. Best thing my first boss ever did for me was give me his office he wasn't using - he'd come in a couple of days a week for 15-20 minutes and we'd talk about things to do, and the rest of the time I'd just do them - and for 6 or so years people left me alone to get things done, some days with 2 interruptions or less. Even working from home, those days are long since over. I'm not "into LLMs" because I am afraid of irrelevance in the next 5-10 years, though - I do potentially want to work for 15-20 more years if they continue to pay me like they do now for the level of engagement I give them, and LLMs or whatever follows them do appear to be likely to be a huge part of that future.

              --
              🌻🌻🌻🌻✌️ [google.com]
          • (Score: 3, Interesting) by turgid on Monday October 05, @01:14PM

            by turgid (4318) Subscriber Badge on Monday October 05, @01:14PM (#1455719) Journal

            I took a different approach. Many years ago I set about implementing simple tools to help me write code faster and more accurately. There will come a point when LLMs become more useful to me, but not yet.

        • (Score: 2) by JoeMerchant on Monday October 05, @09:33PM

          by JoeMerchant (3937) on Monday October 05, @09:33PM (#1455787) Journal

          Tell me you haven't used LLMs without saying "I haven't actually used LLMs":

          > A machine wrote it, it must be right

          Yeah, no. LLMs get stuff wrong about as often as I do, when I'm trying something new to me. The key is to know when you got it right or wrong and not to proceed to build atop that layer until you know the difference. Claude and friends are advancing rapidly in that kind of behavior - writing tests without being told to write tests, even documenting decisions in durable files findable from the code folders.

          One of my main points of discomfort with the LLM generated code is that I think to myself: that's about a 4 line script, and it could be: if you ignore all the error cases, do zero input sanitization, don't carry any internal documentation / help / version info, just assume only the happy path will ever happen. So, lately Claude will come back with 200 lines of all the stuff that I would probably get around to eventually if this is to be a maintained production piece of code, except I'm not used to starting there, I usually add the other stuff when needs are demonstrated one way or another. Does Claude Opus 5.5 get all 200 lines correct on the first try? Occasionally. Whether it does or not, the "right way" to make my ersatz 4 line script properly is to scaffold around it with another 100-300 lines of automated test code, ensuring it works as expected for both the happy path and all anticipatable non-happy cases. Is 500 lines of code, comments and documentation insanity for something that "could be done in 4 lines"? That depends, does anything ever go wrong in your world?

          Then you need another ersatz 4 line script, which you could make "maintainable" with another 500 lines of code, but if you're clever about it, those two scripts can share common modules and trim their resulting code from 1000 lines down to maybe 700-800 lines, and this kind of thing snowballs as you start having 15-20 of these "4 line scripts" which share all kinds of common boilerplate. That's architecture, and if you take the time to push the LLMs to build it, they will, shrinking what could have been 10,000 lines of code down to a couple thousand, including some pretty .svg diagrams showing how the common stuff is shared.

          > here are some superficial unit tests to placate you.

          That was very true in mid 2025, they have significantly stepped up the testing and coverage of regressions game since then, without you having to run around behind them and chide them for strawman tests.

          > Language? Everything's Python these days anyway...

          I did a calendar app using LLM agents in Python, it was... acceptable, but a little laggy. I also was highly disciplined with making sure the code and documentation, requirements specs and tests were good all through the Python development. I asked for strategies to make it faster, port to Go was the path chosen - it did that 100% automatically, all I did was enter "continue" every so often and after I forget how long, maybe 6 hours?, there was the Go app performing identically to the Python app, serving the same database to the same http server with identical rendering, just a lot faster - even served from a Raspberry Pi 3B. I have about 2 hours of prior experience with Go in the past 20 years.

          I wouldn't trust it in "100% self driving mode" for anything less well defined than that language port, and even then it needs the human tests to ensure it really worked, but consider for a moment that it also wrote all that documentation (based on chat input) that made the port go flawlessly...

          > Try not to take it personally.

          This isn't personal, but it is already influencing business bigly and will continue to do so in the future.

          --
          🌻🌻🌻🌻✌️ [google.com]
      • (Score: 2, Insightful) by Type44Q on Monday October 05, @08:31PM

        by Type44Q (4347) on Monday October 05, @08:31PM (#1455780)

        That's certainly one way to reveal that you struggle with analogies...

    • (Score: 3, Funny) by jb on Wednesday October 07, @05:22AM

      by jb (338) on Wednesday October 07, @05:22AM (#1455939)

      Go back in the git cvs history to before LLMs were git was invented. Fork. Leave the loonies to their madhouses.

      That's probably a better proxy for finding "the time when cargo cults took over".

  • (Score: 4, Insightful) by hunger on Monday October 05, @09:55AM (2 children)

    by hunger (4903) on Monday October 05, @09:55AM (#1455701)

    Someone gets kicked out of all communities they want to be a part off? Maybe they are an annoying asshat full of conspiracy theories everybody else is tiered of hearing?

    That post definitely reads like that to me.

  • (Score: 3, Insightful) by Anonymous Coward on Monday October 05, @11:13AM (6 children)

    by Anonymous Coward on Monday October 05, @11:13AM (#1455707)

    What a gut-punch.

    It is extremely believable though.

    Red Hat, Canonical, Continuum et al have always been pirates.

    If you want something free that eventually just needs 'help' which is available at the low low cost of a continuous subscription model paying forever, then buy from a company.

    If you want good free open source software, you've gotta run the systems that are actually actively maintained by the same people who use them.
    They will also be doing the lions share of the work for the companies in category one above there, btw. You'll just be getting a better experiance getting things fixed very quickly, because you live 'closer to the devs'.

    Apple has turned into Microsoft, and Microsoft is seemingly doing its level best to kill itself. Apple is dying slowly too: They've spent too many product refreshes engineering antifeatures for disposability. They probably have the corpse of Steve Jobs hooked up to an alternator, so they can reap free power from him spinning so much in his grave.

    Linus has a great big blind spot: He wants to think he does 'only a kernel', and he's let corporate interests slowly effect a EEE all around the Linux kernel. And that system? It's all the OS stuff that "doesn't belong in a kernel".

    Linux totally should have a non-kernel-mode thingo that does what systemd does, but better. But it doesn't, because that's 'user mode'.

    There are corporate interests best served by the kind of behaviour outed in this article.
    Would MSFT fund such a thing, would Red Hat? Hell yes they would, and you know it.
    That is reason enough to assign a fair quantity of credibility to it.

    Look at the first few comments here? Supportive? Not at all. 'Fake news'.

    • (Score: 2) by JoeMerchant on Monday October 05, @11:47AM

      by JoeMerchant (3937) on Monday October 05, @11:47AM (#1455710) Journal

      This feels like a transplant of political conspiracy theories about "all the major reddit forums have been taken over by Russian trolls." It may be true to some extent, but however true it is, or is not, the Russian human population stands somewhere less than 150M. They may be using bot farms and all kinds of "amplification weaponry" but they don't have access to any more of that than the 450M residents of the EU, or the 340M residents of the US. Now, if China and India in the game and playing hard (which, historically, they haven't - but the internet gives them "proximity" they've never had before...) that makes an interesting evaluation.

      Is turnabout fair play? The colonized having a turn as colonizers?

      --
      🌻🌻🌻🌻✌️ [google.com]
    • (Score: 4, Touché) by JoeMerchant on Monday October 05, @12:13PM (4 children)

      by JoeMerchant (3937) on Monday October 05, @12:13PM (#1455712) Journal

      >Linux totally should have a non-kernel-mode thingo that does what systemd does, but better.

      Personally, I respect Linus for his scope control discipline - the kernel is more than big enough for one "team lead" to handle, git was a kernel development tool that others found useful.

      Love him, hate him, that systemd guy who-shall-not-be-named managed to develop something that has been adopted by the real-world environment, he doesn't have a gun to anybody's head, there are many independent adopters of systemd and they have chosen it over Sys V Init or any limp startup also-rans. You may say the decision makers who chose systemd have a gun to your head, and that's true, but nobody is stopping anybody from duplicating their organizations and running a systemd free alternative. I see a similar, perhaps slightly less contentious, fight looming over Wayland replacing X - the only value X has for me is all the learning I have done about ways to mitigate, incompletely, its insecure architecture. NVidia just did a performance test and found the latest versions of Wayland to be on-par with X on modern hardware, but still people will complain because it's different. Change is unpleasant.

      I find that the obscurity and non-intuitiveness of systemd leaves me frequently choosing to operate in userspace - post desktop load autolaunches instead of systemd config files, because the systemd units are too painful to debug and too often lead to unpredicted outcomes when circumstances change from the anticipated test cases. Full userspace application startup is far more predictable, for me - and the people I work with. Even still, by now a lot of people have learned hard lessons about systemd configuration, and if you're building a "more friendly" alternative you're going to encounter adoption pushback from that, now largely silent, part of the community.

      --
      🌻🌻🌻🌻✌️ [google.com]
      • (Score: 2) by Thexalon on Monday October 05, @03:18PM (1 child)

        by Thexalon (636) on Monday October 05, @03:18PM (#1455731)

        I've been looking into what I think needs to happen about this, and I think it's going to first involve documenting clear protocols that applications use to communicate with systemd. Only then could you replace the systemd side of the equation with something else that does the same basic job in a different way.

        --
        "Think of how stupid the average person is. Then realize half of 'em are stupider than that." - George Carlin
        • (Score: 2) by JoeMerchant on Monday October 05, @04:51PM

          by JoeMerchant (3937) on Monday October 05, @04:51PM (#1455745) Journal

          Yeah, my biggest systemd gotcha so far involved a "dependsOn" that unexpectedly caused a restart loop in something that, ideally, would wait for the thing that it "dependsOn" to be running before starting, but in the absence of it should just start anyway and complain, and in no circumstance would I expect a failure of the dependent process to trigger a restart of the thing that it dependsOn when that thing is running normally - but that's what happened, so a driver failure for this sub-process caused the main message broker for the system to restart loop every 5 seconds, killing functionality of the rest of the system. And it wasn't easy to unwind and figure out what was happening.

          Rather than replacing systemd, I think the most successful approach would be to make it more easily understood and improve visibility of "gotchas" like that. In short: evolve systemd back to have what Sys V Init had in the first place, at least the good parts.

          --
          🌻🌻🌻🌻✌️ [google.com]
      • (Score: 3, Interesting) by Rich on Monday October 05, @11:27PM (1 child)

        by Rich (945) on Monday October 05, @11:27PM (#1455794) Journal

        Corporate vs. Volunteers? Corporate always wins. So RedHat decided they want a clone of Apple's LaunchD. If the Debians won't integrate it themselves, a few breadcrumbs fall towards the Debian startup maintainers to "build a package". Soon these will be happy to include systemd, be it because they got used to it, or just want future contracts. Down the road, RedHats adds whatever they want to systemd, giving us the awful non-modular design we have now - they can just use the lever of "updates". The core idea of systemd isn't evil. Just how it grows into a monolith and taints everything it touches is. Still, it caters to most people's need.

        Wayland is another story.

        First, Wayland will not reach back to allow what's known from broader desktop computing. Applications are not able to order or position windows, by political decisions of the makers. If you see this happen on a Wayland platform now, it's because the applications really use the X11 protocol into a Wayland compositor. There are several proposals and rogue solutions, but the fragmentation with the compositor zoo forces a least common denominator approach for any software to be widely used, mutilating it into a smartphone app. For full desktop behaviour on "pure" Wayland, you'll have to implement your own "Windows MDI" look.

        Second, Wayland can't scale down to lesser hardware. Again, this is a blind "do it like Apple" approach for the wrong reasons. The entire protocol relies on the client pushing updates into a server side bitmap store per window, "Amiga Superbitmap"-style . There is no redraw region handling, therefore making it impossible to run on lesser hardware that just has memory for a plain framebuffer and/or no hardware suitable for performant compositing.

        In general (I have to work with GTK right now) I find that the software design quality in the Linux world has gone down the drain since about 2008. Too many powertripping mental sickos.

  • (Score: 4, Interesting) by VLM on Monday October 05, @01:46PM (1 child)

    by VLM (445) on Monday October 05, @01:46PM (#1455724)

    The actual code, developers and mailing lists are still good. But pretty much everything around it like the IRC channels are getting flooded with people that do the things I've mentioned above

    This is NOT new its always been the case going back to the 90s that there's the people who "do" and the people who "do community" and the paths don't cross much and the "community organizer" types have always seen their role first as political officer / KGB agent than a distant second as friendly open minded marketing people.

    A tale as old as time. One could almost say that people who "do" attract a cloud around them of people who "don't"

    • (Score: 5, Insightful) by ShovelOperator1 on Monday October 05, @08:00PM

      by ShovelOperator1 (18058) on Monday October 05, @08:00PM (#1455776)

      It should be noted that even in the worst "Eternal September" period it was a bunch of trolls, while today there is a coordinated group of people who get paid to do this. As they got the Internet for granted, and it's the corporation from who they get benefits, this cannot be fixed by rules and FAQs.
      And in fact, in current economy regime, cannot be fixed at all, as the economy of destroying whatever cannot be stolen and sold is the economy working as intended.

  • (Score: 4, Interesting) by bzipitidoo on Monday October 05, @05:18PM (7 children)

    by bzipitidoo (4388) on Monday October 05, @05:18PM (#1455752) Journal

    I have long been inclined to think that fresh starts should be tried more often. How did we go from 16M RAM being a generous amount of memory (640K is enough for anyone?), to the current situation in which 16G is now merely adequate?

    I realize much of that is genuine progress. Can't upgrade from ASCII to UTF-8, so that all kinds of symbols (Greek letters, mathematical notations, and even emojis) can be casually used, without needing lots more memory. And, moving from 8bit to 16,32, and finally 64bit blows up the space requirements of something as fundamental as memory addressing, and that percolates all through software. A huge gobbler of resources is of course graphics. Moving from text terminals to GUIs, yes, very good, but it was expensive.

    Yet when I see the average Android app clocking in at tens to hundreds of megabytes in size, I can't help wondering why they are all so fracking big, and suspecting huge amounts of bloat. Is 90% of a typical app graphics data? Is it a couple dozen layers of libraries forming a big pyramid of over 100 libraries?

    The modern web browser has to handle all kinds of standards that didn't exist in the 1990s, while still handling the formats from those days such as MP3, JPEG, PNG, and GIF. Now there are audio codecs such as Opus, video codecs such as MPEG-4, VP8, VP9, and AV1, and for images WEBP, AVIF, and soon, JPEG XL. HTML is now on version 5, and it includes such things as SVG and Canvas. The browser has to compile/interpret/run a full blown programming language, JavaScript, on the fly. XML has been criticized for bloat, can't we improve on it, say, by switching to YAML? Work on Markup to converge on a standard that is superior to the many dialects currently in use? Another big standard is PDF, which browsers didn't do, until this decade. All that is a lot of library code that may not be in the OS, or if it is, may not be good enough quality, and so must be bundled with the browser.

    A reboot could clear out a lot of cruft. Can GIF finally, finally be sunsetted? Can we tackle the inherent bloat of having multiple versions of multiple libraries that all do the same thing? For example, the PDF standard has a section for vector graphics, and HTML has Scalable Vector Graphics (SVG). Can't these be unified somehow?

    • (Score: 2) by JoeMerchant on Monday October 05, @09:37PM

      by JoeMerchant (3937) on Monday October 05, @09:37PM (#1455788) Journal

      >(640K is enough for anyone?)

      1987 I was splitting a DOT spreadsheet across multiple files and chaining results from one to the other because it was too big to fit in available (640K) RAM... no, that was not enough, even then.

      --
      🌻🌻🌻🌻✌️ [google.com]
    • (Score: 2) by JoeMerchant on Monday October 05, @09:39PM (2 children)

      by JoeMerchant (3937) on Monday October 05, @09:39PM (#1455791) Journal

      > upgrade from ASCII to UTF-8, so that all kinds of symbols (Greek letters, mathematical notations, and even emojis) can be casually used, without needing lots more memory

      Ummm.... UTF-8 uses 3 bytes per character MAX, and most common characters (in USian English anyway) are still 1 character ASCII compatible, and most of the rest that are commonly used are 2 bytes. Now, the font files to represent all that do take more space, but do they really need to all be loaded into memory at once? There's only space for so many characters on the screen at one time, why would you waste RAM on loading a bunch of glyphs you're not displaying? ;-P

      --
      🌻🌻🌻🌻✌️ [google.com]
      • (Score: 2) by bzipitidoo on Tuesday October 06, @03:12PM (1 child)

        by bzipitidoo (4388) on Tuesday October 06, @03:12PM (#1455859) Journal

        > but do they really need to all be loaded into memory at once?

        Yes, yes, they do! Fast Start(up) and similar things, you know....

        • (Score: 2) by JoeMerchant on Tuesday October 06, @04:42PM

          by JoeMerchant (3937) on Tuesday October 06, @04:42PM (#1455865) Journal

          Back in the day of 88kB floppy discs I had a plan to make a controller for a 5MB hard drive to emulate a collection of 55 floppies - of course things were changing so fast back then and design was so much more difficult, native hard drive support arrived before any such beast was actually created, but... in the 6502 "64kB of address space is all you need" world, you could do "page-in" tricks where you could have a massive ROM store that gets selected in as an 8 or 16Kbyte page of the 64K address space, rapidly build the special characters table you need by copying the glyphs from ROM to RAM while building a lookup table, in about the time it takes a floppy disc to rotate one revolution, or less...

          --
          🌻🌻🌻🌻✌️ [google.com]
    • (Score: 3, Interesting) by arslan on Tuesday October 06, @02:44AM

      by arslan (3462) on Tuesday October 06, @02:44AM (#1455807)

      My take on bloat:

      - lots of bundled tracking/ad ware for monetization
      - tree shaking (and related minification/compression) seems like an afterthought in a lot of platforms/frameworks/languages, which seems odd given we're in 2026
      - lots of bundled tracking/ad ware for monetization
      - unnecessary meta-data in media files, likely for tracking
      - lots of bundled tracking/ad ware for monetization
      - your note on catering to a circus of formats
      - lots of bundled tracking/ad ware for monetization

    • (Score: 1) by shrewdsheep on Tuesday October 06, @07:47AM

      by shrewdsheep (5215) on Tuesday October 06, @07:47AM (#1455820) Journal

      And, moving from 8bit to 16,32, and finally 64bit blows up the space requirements of something as fundamental as memory addressing, and that percolates all through software.

      Linux had (has?) the (compile time) mode of using 64bit for data and 32bit for addressing. It was measured that the 64bit vs 32bit addressing comparison implies a rather modest overhead (5% if I recall correctly), so blowing up the address space does not seem to be a substantial penalty. (recalling this off the top of my head, sorry no citation).

    • (Score: 0) by Anonymous Coward on Tuesday October 06, @09:33AM

      by Anonymous Coward on Tuesday October 06, @09:33AM (#1455830)
      Well nowadays a 4k display requires at least 24MB if not 30MB.

      Then caching stuff for alt-tab takes more MBs.

      Then caching some data for back/forward button for similar reasons.

      Then say you have 10 different tabs open...

      Then suddenly hundreds of MBs used seems not that sub-optimum compared to rebuilding and redrawing everything from scratch every time.

      Add web pages with complicated JS, and lots deciding to rewrite desktop apps as web apps and you get GBs used... 🤣
  • (Score: 4, Interesting) by The Vocal Minority on Tuesday October 06, @09:07AM (4 children)

    by The Vocal Minority (2765) on Tuesday October 06, @09:07AM (#1455827) Journal

    Very interesting and worrying article. I don't know enough about the Gentoo project to have a sense of how accurate the linked comment is, and I'm one of the more pro AI people here so that aspect doesn't bother me so much, but corporate bullying via CoC does. I've seen it done on a smaller scale so I can just imagine it happening here. Any comments from people closer to this who know what's going on?

    • (Score: 2) by bzipitidoo on Tuesday October 06, @03:34PM (3 children)

      by bzipitidoo (4388) on Tuesday October 06, @03:34PM (#1455860) Journal

      I've been running Gentoo Linux for the past 3 years now, because in Gentoo, systemd is optional, and I opt not to have it.

      However, Gentoo still takes way too much time to compile everything. I had foolish hopes that speed improvements, particularly the SSD, would reduce compile times so low it would barely register. Nope! Still takes hours to do an update. Next time I upgrade, I'm changing distros. Again. The hunt for the best distro never ends.

      • (Score: 2) by JoeMerchant on Tuesday October 06, @04:52PM (2 children)

        by JoeMerchant (3937) on Tuesday October 06, @04:52PM (#1455868) Journal

        What's a full system bring-up in Gentoo these days?

        Last time I did one (20-ish years ago) it was right around 24 hours.

        --
        🌻🌻🌻🌻✌️ [google.com]
        • (Score: 3, Informative) by bzipitidoo on Wednesday October 07, @12:20AM (1 child)

          by bzipitidoo (4388) on Wednesday October 07, @12:20AM (#1455919) Journal

          Yeah, 24 hours is still about average, but it is highly variable. First, the machine has to have good cooling. The computers I've had over the past decade are too prone to overheating when under the sustained load of doing hours of compiling. I've had to use various measures to deal with that. I hacked up a quick and dirty pause for the compiling, just a dumb script that does "pkill -STOP emerge;sleep 90;pkill -CONT emerge; sleep 30" endlessly.

          The most brutal compiles are the big programs such as the browsers and big libraries such as webkit. Those take around 2 hours each. Slowing the compiles down like I did drags that out to 8 hours each, but slowing it down that much means I can use it while it updates. Gentoo does not make it easy to continue mid build, generally have to start over the build on whatever package it was on, which is not an issue for a small package, but if it overheats and locks up near the end of building a big package such as Chromium (~54000 steps!), it's a bummer. That's with my desktop, an AMD Ryzen 5600G with a nice, big, fast SSD and 32G RAM.

          One consequence of such a heavy update process is being reluctant to get every Firefox update. I often skip a version or two.

          • (Score: 2) by JoeMerchant on Wednesday October 07, @12:44AM

            by JoeMerchant (3937) on Wednesday October 07, @12:44AM (#1455920) Journal

            >the machine has to have good cooling

            My Gentoo machine from circa 2005 was an AMD 64 bit, back when AMD were the low power FLOPS per watt masters of the universe. Excess heat never was an issue for that full sized tower.

            I migrated from Gentoo to Ubuntu circa 2008/9 IIRC, and since 2024/5 I've been choosing Debian over Ubuntu when there's a choice. Glad to see that all my reasons for leaving Gentoo behind are still valid, thanks!

            If you maintained a "fleet" of Gentoo machines could you do your custom builds on a single server and push the binaries out to the rest of the systems, or is that not a supported concept? I'm having a sick vision of a little under-powered over-cooled compiling hub constantly building updates for the rest of the local network.

            --
            🌻🌻🌻🌻✌️ [google.com]
(1)