2026-03-04 - This post was originally shared over the Gemini protocol on my gemlog there, on 2021-11-14. It was a response to a chain of gemlog posts titled “How you were using the Internet in 1991-1995 and 1995-2005?”. It was meant to be part one of a two-part response, but I never got around to making the second. Maybe I will do that soon, though.

1998-2000

I was born in 1988, so I was too young during 1991-1995. My first time using the Internet was in 1998. We had the AOL suite on a dial-up modem that used our landline. My screenname was BBJeff10 – BB for baseball, my favorite sport, and 10 because that was my age at the time. My mom set up our screennames, and her user was the master account on our AOL instance. I don’t quite recall what I did on the Internet exactly, but I didn’t use it a whole lot. I think I mostly just chatted with my friends on AOL and looked up video game stuff on GameFAQs and Cheat.cc.

2000-2001

I was really into professional wrestling entertainment, like WWF (now WWE), WCW, and ECW. I found these fantasy wrestling communities called e-federations or e-wrestling leagues. These were communities of users who create their own original characters and roleplay various scenes, challenging other wrestlers in the league. The roleplays were written out and posted on the league’s message board. Feuds between wrestlers would take place, and if there was a match between them, the wrestler with the best roleplay would win, as judged by the league owner/CEO who was usually the webmaster and did not take part in roleplays. The matches were simulated using wrestling simulation software. One such program in particular was called Zeus Pro.

The league that I was part of was called ACWL (Action Championship Wrestling League) and the owner, who was called Jim Rooster, wrote his own wrestling simulator software. My character had a caustic and cocky personality and his wrestling style was a mix of lightweight/high-flying and hardcore/roughneck. My character was inspired by the urban grunge/gothic aesthetic of the wrestler Raven, the hardcore dare-devil-ish Cactus Jack/Mankind, and the martial arts of Steve Blackman. He had some crazy high-flying moves too, like shooting star press and moonsaults.

I made one online friend within my first week of roleplaying; one user liked my roleplays and my character and asked me if I wanted to team up with their character. Their character was named Rune, and we formed a group called the Renegades of Funk. Our theme song was something by Rage Against the Machine (it might have been Guerilla Radio, but I can’t say for sure. Or that might have been my own character’s theme song.) Our partnership didn’t last very long but I don’t remember why, I think we just decided to go separate ways.

The owner/CEO of the league deemed me the most creative trash-talker in roleplays; my character won the ACWL North American title belt and held onto it for two matches before losing it to a character named Cyborg, with whom my character feuded for a time. Some users’ roleplays were kind of perfunctory and simplistic; mine were among the more detailed ones.

Anyway, during this time I became more computer/Internet savvy because I spent a lot more time with it. I made my AOL username the master account, so I was able to view websites for wrestling news and M-rated video game stuff that was restricted by PG-13 AOL parental controls. HTML was often used for roleplaying on the message boards, and eventually I realized I could use HTML to design a whole website. At one time I had my own e-federation with a website I designed that was hosted on Angelfire. My e-fed didn’t get very big, there were maybe 10 or so people, most of whom were online friends I made from the other league.

During this time also was when Napster was really popular. I thought it was amazing that I could look up any individual song and download it, for free, to my computer.

I will follow up on this post and share my Internet experience from the years 2002-2005.

2026-01-06 - IMPORTANT UPDATE: I have learned more about the schizoid temperament from Nancy Mcwilliams’ paper, and I now believe that I am in fact a schizoid. My conclusion, and my framing of the schizoid temperament in this post in terms of its pathologized DSM criteria, are, in my opinion, obsolete.

I recently mentioned having a “schizoid temperament”, and I’m wondering to what extent my temperament is genuinely schizoid, or if it can be explained by other things.

According to sources on Wikipedia, schizoid personality disorder “is characterized by a lack of interest in social relationships”, also known as asociality, “a tendency toward a solitary or sheltered lifestyle, reservedness, emotional coldness, detachment, and apathy.” I’m going to unpack what this means and whether or not I fit the bill.

Asociality

I don’t have a lack of interest in social relationships. I actively seek out and interact with like-minded people online. I don’t do this in person. In person, I’m reserved, standoffish, I avoid social interaction, and I prefer to be alone doing solitary activities. I don’t feel comfortable in social settings unless I’m with family. From self-observation, the primary reason for this is because I don’t see much of a point in socializing, small talk, or potentially putting myself into anxiety-inducing situations. I’ve kind of implicitly learned that most people I encounter in person will not share my values, beliefs, and interests, and this is a source of social anxiety, feeling disconnected, and lack of interest. This is true even among family, so I tend to avoid my special interest topics and instead just go with the flow of the room conversation. An example justification for this that isn’t just hypothetical: A family member asked about an EFF sticker on the back of my car. EFF is the Electronic Frontier Foundation, and they focus on privacy and digital rights for consumers of technology. I wasn’t able to explain this with speech, and what came out of my mouth was “it’s a computer thing”, hoping the subject would change or that it will be enough for them to not ask further questions. My “obsessions”, or special interests as they are called, often make it hard for me to stay even at family gatherings, though I do try to hold out for the sake of my family. I know they would prefer me to be there, and I think family gatherings are good for a healthy family dynamic. I’m fortunate to not have a dysfunctional family where people don’t speak to or see each other. But my special interests call to me, so I end up usually leaving early.

I think any asociality I exhibit is primarily due to social anxiety and introversion, or possibly autism. I have no interest in changing this behavior, and I don’t think it has an adverse effect on my general psychological well-being. I’m perfectly happy doing solitary activities and interacting with people online. It’s weird how other people will typical-mind and project their own values onto this sort of lifestyle, for example by saying things like “I’d be so unhappy if I was stuck at home all the time, you gotta get out and do things and socialize.” It’s weird because their intention is to be empathetic, e.g., “how would I feel in his situation?”, but what they end up doing is imagining my situation from the context of their own values and interests. If you’re extraverted, then of course you’d be unhappy in a solitary, sheltered lifestyle. They mistake their own values and personality preferences as being what all humans should aspire to.

Emotional coldness, detachment, and apathy

I wouldn’t describe myself as emotionally cold and detached. At least I don’t feel that way when I’m at home in my comfort zone. However, in person, I do often feel disconnected from others, and this may manifest as standoffishness and apathy. I’m friendly and cordial with strangers in public, even when I harbor paranoid suspicions about them. The paranoid suspicions I have don’t tend to be things that would make it difficult to be friendly and cordial, and even when they are, I’m usually theoretically aware of the difference between my brain and the external world, so I try not to let my suspicions govern my behavior, nor do I like drawing negative attention to myself.

The Wikiepdai article for schizoid personality disorder mentions an association with dismissive-avoidant attachment style. My attachment style is more anxious-avoidant, which fits with the theory that any schizoid traits I exhibit can be explained better by social anxiety, agoraphobia, and introversion – or, again, possibly autism.

Another thing mentioned in the Wikipedia article is that schizoid people feel like they are detached observers of their own behavior. Big check mark for this one, but I think that’s just where my paranoia overlaps. This is also not exclusively a schizoid trait.

In the early days of my mental illness, I used to have what I would describe as “delusionally-influenced stilted speech”. This is no longer the case today, or at least as far as I’m aware, and is probably due in large part to medication, and in no small part to gaining social experience, and social and emotional maturity.

I cancelled my $9.99/month Door Dash subscription, and am either going to buy food from the grocery store, or, if convenience is more warranted, leave the house to get food. Door Dash delivery is effing expensive.

I’m also going to downgrade my internet from Comcast. I currently pay $108 per month for 2.2 Gbps speeds. I can get reasonably fast speeds, like 1000-1200 Mbps, at roughly half that price. I anticipate that the benefits I currently receive from a 2.2 Gbps internet connection will not be terribly or even noticeably reduced on a 1000 or 1200 Mbps connection. As an aside: For the first two years I lived with my grandpa, I was making due with 12 Mbps Internet from his Comcast account, and that was fine for what I was using the internet for at the time.

I started making a recurring monthly donation of $3.10 to the Wikimedia Foundation. I might also start a $5 per month donation to the Internet Archive.

My VPS and cloud server expenses are exactly what I want them to be, for now. I’m also thinking of not renewing my hyperreal.coffee domain, and just using my moonshadow.dev domain. The only service using hyperreal.coffee is my web server. Most other things use moonshadow.dev, including my email address. I see little reason to spend money on two domains, and the easiest one to cut is hyperreal.coffee. I’ve had it for several years, and it was fun while it lasted, but I don’t have a hard need for it anymore. Any links that break because of this are not terribly important or mission-critical. The same contents of my blog and website will still be available at moonshadow.

For the past year, I was using Fastmail as my email service. My subscription of $60/year is due to expire in the next month, so I cancelled it. I’m now back to using Proton Mail with a Proton Unlimited plan for $9.99/month, which includes a ton of other stuff besides email. There is not yet a Proton Drive client for Linux, and I don’t anticipate there will be one anytime soon, but my Nextcloud with Murena Workspace setup is working just fine for now. Since I have 500 GB of storage space on Proton Drive, I’m going to write a script that rclones snapshots of my Nextcloud directory to my Proton Drive account, to use as an alternate backup source for my Nextcloud content.

Speaking of the Murena Workspace account. It’s nice to have someone else managing it, but that is another $2.33/month expense. I’m considering canceling it. All I really need out of it is a file sync. It might be feasible to write an rclone script with functions to push and pull content to my Proton Drive at regular intervals. On my desktop machine, the idea would be to run a push function every hour, and only run the pull function manually when I use another device like my laptop. I can accomplish a similar functionality using external drives and ZFS send/receive, but since 500 GB of storage is already included with my Proton Unlimited plan, I might as well make use of it and benefit from the offsite storage.

I struggle finding a goldilocks distro because my personal constraints are variable at arbitrary times, and sometimes these constraints arise after some negative experience with a distro, even a relatively minor one that interrupts my activity flow. Then I have to do some weird calculus or non-linear programming (emphasis on the non-linear) and may end up back where I started.

The problem is never completely solved. It may be partially solved temporarily and I just put up with certain constraints not being satisfied until I don’t want to put up with it anymore. It’s frustrating, and I realize how ridiculous it is given that there are far more important and consequential things in the world to be frustrated about.

Add to this my ADHD-like whim following when I see someone posting about a distro or something on social media and I’m like SQUIRREL! The fuck, man.

It’s such a shame that it’s partly proprietary. But I don’t know. It’s tempting for me to be open-minded about using it.

• 92% of the code base is open source from Chromium.
• 3% is open source from Vivaldi.
• 5% is proprietary from Vivaldi, which is mostly the UI stuff.

They’ve also made a public commitment to not integrating AI into their browser. This gives them points for trustworthiness and putting the user above AI hype and profit.

So… I’m not usually in the habit of giving hot takes on software, but I’m kind of thinking that despite its 5% proprietary code, and in light of Mozilla’s terribleness lately, it’s ethically and technically superior to Firefox.

A lot of what Shuttleworth said resonated with me – he mentioned he’s fascinated by the small things self-hosting and IoT stuff. He also discussed the merits of Snap packages, which made me less hostile to the fact that Ubuntu prioritizes them over regular .deb packages.

Shuttleworth seems like a pretty cool guy. Hopefully he stays that way.

Another thing that resonated with me in the podcast is that the hosts talk about how the big players in the FOSS ecosystem – Ubuntu/Canonical, Fedora/Red Hat, Suse/Novell – all raise the tide for the smaller community-oriented distros and projects. So it’s like a symbiotic relationship in that regard. It’s not so much a zero-sum game for distros, even though there is some nominal level of competition among them.

I never knew much about him other than that he developed LSD, and I suspect that’s what most people think of when they hear his name. There is quite a legacy behind the person.

The following text is quoted from the Facebook page This Day in History, with my own paragraph spacing added.

https://www.facebook.com/share/p/16jkn5KcEP/

Timothy Leary arrived at California Men’s Colony prison in 1970 facing twenty years behind bars. He was fifty years old, a former Harvard psychologist turned counterculture icon, sentenced for marijuana possession in an era when the establishment treated drug advocacy as existential threat. Prison officials processed him like any other inmate: fingerprints, prison blues, a psychological evaluation to determine security placement.

Then they handed him the test.

Leary looked at the questions and recognized his own work. Years earlier, during his respectable academic career, he’d developed the Leary Interpersonal Behavior Inventory—a personality assessment designed to evaluate how people relate to authority, handle stress, and respond to institutional control. Now the same system he’d created was being used to classify him.

He filled out the questionnaire in under ten minutes, deliberately shaping his answers to present exactly what the evaluators wanted to see: a middle-aged man with no fight left, interested in gardening and forestry, compliant, unlikely to cause trouble. The results placed him in minimum security with an assignment tending the prison gardens. Leary had just manipulated his own psychological test to engineer the conditions for his escape.

He wasn’t always a rebel. Timothy Leary started as an establishment academic, exactly the kind of psychologist Harvard hired in 1959 expecting orderly research and publishable papers. He’d earned his PhD from Berkeley, directed psychiatric research at Kaiser Foundation Hospital, published respected work on personality assessment. His career trajectory pointed toward tenure and professional respectability.

Then he went to Mexico in the summer of 1960 and tried psilocybin mushrooms.

The experience didn’t just alter his consciousness—it altered his understanding of what psychology could be. He returned to Harvard convinced that psychedelic compounds offered therapeutic potential that traditional psychiatry couldn’t match. He started the Harvard Psilocybin Project, conducting research with graduate students, prisoners, and intellectuals. Subjects included Allen Ginsberg, Jack Kerouac, and religious scholars. They took controlled doses, wrote detailed reports, participated in therapeutic sessions.

The research was legitimate. The methodology was sound. But the implications terrified administrators who saw a professor encouraging students to chemically alter their perception of reality. Colleagues whispered about loss of scientific objectivity. Newspapers sensationalized the experiments. By 1963, Harvard decided Leary had become a liability. They fired him, officially for leaving campus during the semester without authorization, but everyone understood the real reason: he’d stopped being a respectable psychologist and become something the institution couldn’t control.

Leary saw the dismissal as liberation. Freed from academic constraints, he became the public face of psychedelic exploration. He traveled, lectured, founded research centers first in Mexico then in Millbrook, New York. He coined phrases that became counterculture mantras: “Turn on, tune in, drop out.” Critics heard nihilism. Followers heard permission to question inherited assumptions about consciousness, authority, identity.

The establishment heard threat.

In 1965, authorities arrested him at the Texas-Mexico border for marijuana possession. Prosecutors pushed for harsh sentencing, treating him as a symbol of cultural decay that needed crushing. Leary fought back through the legal system. He appealed all the way to the Supreme Court and won—not through cultural arguments about drug policy, but through constitutional law. The Court ruled in Leary v. United States that the marijuana tax law he’d been charged under violated the Fifth Amendment’s protection against self-incrimination.

But victories in court didn’t stop the targeting. In 1968, police arrested him again in Laguna Beach. Combined with his previous offense, he now faced twenty years in prison. This time, there would be no Supreme Court rescue.

Once transferred to the minimum-security facility at California Men’s Colony, Leary began planning the impossible. The Weather Underground, a radical leftist organization, agreed to help for $25,000 paid by The Brotherhood of Eternal Love, a group of psychedelic advocates. They smuggled him tools and coordinated logistics for a nighttime escape.

On September 13, 1970, Leary climbed onto a prison rooftop, pulled himself up a telephone pole, and moved hand-over-hand along a cable spanning the prison yard—over barbed wire, over security perimeter, beyond what guards thought anyone would attempt. He dropped to the road outside where Weather Underground operatives waited.

He left behind his prison clothes and a note that reportedly read: “I declare myself free.”

The escape became international news. Prison officials found themselves explaining how one of America’s most recognizable prisoners had simply climbed out while they watched. Leary fled to Algeria, where he stayed briefly with Black Panther Eldridge Cleaver’s government-in-exile. Then Switzerland. Then Afghanistan. He traveled under assumed names, gave lectures in disguise, wrote manifestos. Intelligence agencies tracked him across continents. He became a fugitive philosophy professor.

The chase ended in 1973 when Afghan authorities, working with American intelligence, arrested him in Kabul. He returned to the United States in handcuffs. Newspapers framed it as law finally catching up with chaos. Leary framed it differently. In interviews, he emphasized that the real threat wasn’t psychedelics—it was the questions they made people ask about who controlled their consciousness and why.

He served additional years in prison, during which he wrote extensively about neurology, consciousness, and human potential. When finally released in 1976, he didn’t retreat into quiet obscurity. Instead, he evolved. He lectured on cyberculture before most people understood what the internet would become. He talked about space colonization, artificial intelligence, consciousness expansion through technology. He refused to become a museum piece from the 1960s, choosing instead to adapt his message to new frontiers.

Critics dismissed him as a relic trying to stay relevant. But Leary understood something fundamental: the questions he’d been asking—about autonomy, consciousness, institutional control, who decides what’s acceptable to think and feel—didn’t stop being important just because the decade changed.

Near the end of his life in the 1990s, still lecturing and writing, someone asked why he’d spent decades provoking institutions that imprisoned him, surveilled him, tried to silence him. Why keep pushing when the cost had been so high?

Leary’s answer captured everything about his approach to authority, consciousness, and control: “The moment you stop questioning, somebody else starts answering for you.”

Because that was the real story of Timothy Leary—not the drugs, not the slogans, not the escapes. It was his refusal to let anyone else define the boundaries of acceptable thought. He’d figured out how to game a psychological test because he understood how institutions try to categorize and control people. He’d escaped from prison because he understood that the only real prison is the one you accept. He’d kept questioning authority because he understood that the alternative wasn’t safety—it was surrender.

The establishment called him dangerous. Leary called them predictable. And in 1970, when they handed him his own test thinking they’d finally contained him, he proved exactly how predictable they were.

Meerkat (meer9) is now my main workstation, with CachyOS KDE Plasma. I’ve also decided to run homelab monitoring stuff like Ntfy, Prometheus, Grafana, Loki, and Promtail with syslog-ng forwarding on it. I don’t quite see a sufficient reason to keep this monitoring stack separate from my workstation. I do everything there anyway, so the services might as well run on localhost. I can still connect to the Grafana dashboard over Tailscale from my laptop.

I played Prey for a couple hours on Bazzite, and somehow the performance of the game felt more efficient than it did on CachyOS. I don’t know why, but it’s alright by me.

On Bazzite, I also have my Nextcloud via Murena Workspace account accessible on it with the Nextcloud desktop client flatpak. Most of my Steam games store saved games in the Steam Cloud, but GOG stores everything under ~/Gaming/gog/. Lutris allows using the native system paths for some games. Like Dishonored stores saved games and settings under ~\Documents\My Games on Windows. The Linux XDG equivalent would be ~/Documents, which for me is a symbolic link to ~/Nextcloud/Documents, so that stuff is synced with Nextcloud. In contrast, Prey stores saved games under ~/Games/gog/prey/drive_c/Users/jas/Saved Games/Arkane Studios/.... For now I just rsynced the saved games there, but I have to figure out an efficient way to sync that directory on Nextcloud. Maybe if I keep the actual files under ~/Nextcloud/Documents/Saved Games and create a symlink to that directory from ~/Games/gog/prey/drive_c/Users/jas/Saved Games. Now that I think of it, I recall there being a program on GitHub called GameImage that might provide a more efficient way to keep gaming stuff synced on Linux. I can’t just sync ~/Games on Nextcloud because Lutris downloads and stores the entire game files there, roughly more than 100 GB of data with the games I currently have installed. Not something I want to sync over Nextcloud lol.

Anyway I’m pretty satisfied with this setup.

In other news, my NAS has to remain shut down until we figure out what is wrong with our house’s electrical wiring. My NAS is the most consistently power-hungry machine in my homelab, so it will be good to keep it off until we fix whatever needs fixing. My parents’ electrician friend is coming this weekend to see what’s wrong. Last week a circuit breaker shorted out, which powers stuff in our garage and outside the house. The wiring in our house is pretty old, and probably wasn’t setup to handle digital age power demands. I’d like for my parents to consider getting solar panels, but that’s too liberal-coded for them and they hastily dismiss the idea without a second thought. We don’t need no goddamn hippie solar panels.

My bedroom is pretty quiet now.

• Orange Pi 5+ running Ubuntu 24.04, serving as a centralized monitoring station with Prometheus, Grafana, Prometheus exporters for qbittorrent and my Tapo TP-Link smartplugs, and syslog-ng forwarding.
• Orange Pi 5+ running DietPi, serving FreshRSS, Nextcloud, Blocky DNS proxy with Unbound recursive DNS resolver.
• System76 Meerkat (meer7) running Debian 13, serving my public web content, including my website and other public-facing web services. I had port forwarding enabled in my router to this machine.
• System76 Thelio Major running FreeBSD 14.3, serving as a NAS.

I’ve rearranged, cleaned up, simplified, and repurposed the first three. I’m no longer port forwarding through my router.

My public stuff is on a Debian 13 VPS from netcup.de that resides in Manassas, Virginia, USA. From there, it can benefit from higher availability and uptime, easier snapshotting and backup of the entire VPS, IPv6, and it closes off an attack vector on my home network.

Since I use Murena Workspace for Nextcloud and other cloud stuff, which integrates nicely with /e/os on my phone, I no longer need to self-host Nextcloud. It is therefore superfluous to have the DietPi machine running. I can just consolidate the Blocky DNS proxy and FreshRSS into the centralized monitoring machine. However, since my System76 Meerkat is now free, I’d prefer to use it for these tasks instead as it is far more powerful. This means I can reduce four machines into two, namely the Orange Pi 5+ machines will be laid off and the Meerkat will replace them. Sad for the Orange Pi 5+ devices, but they’re just machines without feelings, families, or survival needs, so they’ll be fine if a bit dusty.

As it turns out, Fedora 43 was just released today. So I’m going to install Fedora 43 Server on the Meerkat.

Instead of making a justfile, I think it would be better if I wrote up a blog post on how to get this setup on a vanilla Fedora. This way, people reading it are in full control of all the commands that need to run, the repositories that need to be configured, and the Bazzite kernel to install.

The tricky part, and a problem for me to solve, is keeping the Bazzite kernel up to date with each new release. I figure I can host a public yum repository to store the packages. I can write a script that does the following:

• The current kernel version number will be stored in a file, e.g., /etc/current_kernel_version on the yum repo host machine.
• Check for the latest kernel version in the bazzite-org/kernel-bazzite releases.
• Compare this version to the current.
• If it’s newer, download the newer kernels from the bazzite-org/kernel-bazzite repo releases and move them into the yum repo.

Then when a new release is available, it can be upgraded to with dnf update. I have to figure out a way to make it so that it is prioritized over the upstream Fedora kernel.

But, sigh.

I’m still kind of torn between this project and CachyOS. Everything just works the way I want on CachyOS. There is no hard reason for me to use Fedora. The only thing inspiring me to do this that I’m just used to and like the Fedora ecosystem.

I don’t want to use Bazzite anymore – I just don’t like the restrictiveness of the atomic/immutable filesystem. It was a moderate burden that I tolerated for a few months, but it gets in the way a lot.

Sure, there are ways to do the things I want in atomic/immutable distros, but it tends to require some convoluted hack with containers and CI workflows and trial and error that I’d prefer not to spend time on. I feel like in some ways it’s a just nerd harder thing, for a sense of accomplishment for doing something hard that didn’t really need doing in the first place. It feels like building a Rube Goldberg machine. Whereas all it takes on a normal filesystem is a single command or script that, for example, changes something in /usr/share.

I can’t say whether immutable/atomic distros are “the future of Linux” in the sense that non-immutable distros will “die out” eventually. They clearly have a strong foothold in the wider Linux ecosystem, so they will be around for a long time. In that sense, they are a big part of the future of Linux. My fear is that, with enough backing from corporate sponsors and the consequent momentum/pressure this puts on other distros, immutable distros might eventually become the de facto standard for Linux, and render the non-immutable distros niche. But this fear is exactly that, and doesn’t necessarily coincide with reality. There are strong communities behind the likes of Debian, Arch Linux, FreeBSD, et al, which tend to push back against corporate influence.

I can confirm the OS is “blazingly fast”. CachyOS also provides tooling for setting up and optimizing gaming on Linux, which is nice. Since it is based on Arch Linux, I have access to Arch Linux package repositories, including the AUR. This means I don’t need Linuxbrew. I also have my usual Flatpaks.

It’s also nice to be back on a normal, mutable filesystem and not have to use Distrobox. Though, to be fair, that was only a minor inconvenience on Bazzite and I got used to it pretty quickly to where it didn’t get in the way. But still, it’s nice to not have to do that.

I did not really plan this. It just kinda happened today. But I look forward to exploring CachyOS as a daily driver, for a while at least.

The basic ingredients are as follows:

• Private-facing server with adequate storage, accessed only via Tailscale, to serve as a build host
• Docker or Podman
• docker-registry-ui without authentication (because Tailscale)
• Caddy web server to reverse-proxy the container registry
• bluebuild-cli
• Bash script to run the bluebuild program with logging and notifying on build failures
• systemd service and systemd timer to run the Bash script on a schedule
• Cosign package installed on the build host

The container registry is configured to allow pushing without authentication. I don’t need to worry about authentication because it is not open to the public internet. The bluebuild build command is simply supplied with the registry URL on my Tailnet.

Here is the Bash script I wrote to build Bazzite and Bluefin images and push them to the container registry:

#!/usr/bin/env bash

set -x

DATE=$(date '+%Y%m%d_%H%M%S')
LOGFILE="${HOME}/bluebuild-logs/${DATE}.log"
NTFY_SERVER="https://nas-aux.carp-wyvern.ts.net"

if [ ! -d "${HOME}/bluebuild-logs" ]; then
    mkdir -p "${HOME}/bluebuild-logs"
fi

notify_fail() {
    curl \
        -H prio:urgent \
        -H tags:warning \
        -d "${1}: failed to build. See logs for details." \
        "${NTFY_SERVER}/bluebuild"
}

print_header() {
    echo "===================================================================="
    echo "Start build of ${1}"
    echo "Date: $(date '+%Y-%m-%d %H:%M:%S')"
    echo "Commit: $(git log --oneline --no-decorate -n 1)"
    echo "===================================================================="
}

print_footer() {
    echo "===================================================================="
    echo "End build of ${1}"
    echo "Time elapsed: ${2}"
}

if [ ! -f "$(pwd)/cosign.key" ]; then
    echo "cosign.key not found. Copying it from backup."
    cp -fv "${HOME}/.cosign.key" "$(pwd)/cosign.key"
fi

for recipe in $(/bin/ls ./recipes/recipe-*.yml); do
    print_header "$(basename -s .yml "$recipe")" >>"$LOGFILE"
    start_time=$(date '+%s')
    if ! bluebuild build \
        -p \
        --registry aux-remote.carp-wyvern.ts.net \
        -B docker \
        -S cosign \
        "${recipe}" >>"$LOGFILE" 2>&1; then
        notify_fail "$(basename -s .yml "$recipe")"
    fi
    end_time=$(date '+%s')
    runtime=$((end_time - start_time))
    time_elapsed=$(date -d @$runtime -u '+%H:%M:%S')
    print_footer "$(basename -s .yml "$recipe")" "$time_elapsed" >>"$LOGFILE"
done

# vim: ts=4 sts=4 sw=4 et ai ft=bash

Source: https://tildegit.org/hyperreal/bluebuild-hyperreal

This script is then run with a systemd service and timer every day at 01:00 (CST). In terms of results, it is very similar to running it in a CI/CD pipeline. Each recipe is slated to run, and if one fails, a notification is sent to my ntfy server (also on my Tailnet), and the script continues to build the next recipe. Build output is redirected to logfiles. Everything is run as an unprivileged user from the repo’s root as the working directory.

git pull is run by systemd before the execution of the script, so when updates are pushed to the repo, they are reflected in the next scheduled build. If I need to build them immediately, I can just run systemctl --user start bluebuild_build.service manually to trigger a build. Nothing in the script changes the state of the git repository during the build process, so each run will be ensured a clean git repo.

So, yeah, that’s it. I think it’s kinda neat.

I’m thinking of doing the building on my own machine and pushing the image to a self-hosted container registry. The bluebuild cli makes this easier. For a self-hosted container registry I can use Harbor.

I would write a script that fetches the latest version from my Codeberg repo, runs something like bluebuild build -v --push --registry $HARBOR_REGISTRY recipes/recipe.yml locally, and sends the log output to a file for debugging purposes.

I would have the script run as a systemd service with a daily timer. If I use a systemd service then I wouldn’t necessarily have to store the log output to a file, as I could view it with journalctl -xeu bluebuild-build.service, but it might be nice to have separate log files in addition to that.

This way I can self-host it without having to fiddle with Gitea or Codeberg CI.

The problem with (1) is that I’d like to be able to have one set of peripheral devices that can connect to both my gaming PC and Linux workstation, and conveniently switch between them when I want to. One way this is possible is just by unplugging and plugging in each peripheral device (keyboard, mouse, monitor) into the desired machine. But this would be a pain in the fundament, because my gaming PC doesn’t have any ports conveniently on the front of the machine, so I’d have to crawl down under the desk and reach into the back of the machine every time. I could also use USB hubs for this and save myself from having to crawl under the desk. I can have a USB hub for each machine. This would at least cover the keyboard and mouse. As for the monitor, my current monitor only has one DisplayPort port, so I’d have to keep unplugging and plugging in DisplayPort cables for both machines. Maybe I can find a monitor that has more than one DisplayPort, and use the monitor’s buttons to toggle between them. Or some kind of splitter/switch hub thingy that has a button that I can just press to toggle which machine uses the monitor (if such a thing exists?). Another way would be Bluetooth peripheral devices that I’d have to keep charged and paired constantly – but that idea can go straight the fuck to Hell where it belongs.

What’s wrong with using the Windows 11 gaming PC as my main workstation like a normie instead of having a separate Linux workstation? Honestly, I’ve seriously considered this, as it would sure make all this easier. But the answer ought to be obvious to anyone who cares about digital rights[1]. Why don’t I just dual boot on the gaming PC, then? Because reasons.

[1] Maybe there is a way I can isolate some things similar to Qubes OS, and try to keep personal data off the Windows host OS as much as possible. Maybe I can leverage WSL for this somehow. I can look into making Windows less privacy-invasive without breaking certain functionality, if that’s even possible these days.

Of course, none of this would be a problem if I had more space in my bedroom/office with outlets at convenient locations, but that’s not possible for the foreseeable future, so I have to work with what I’ve got. I think any arrangement I come up with is going to have an entropistic Cyberpunk 2077 look, and, uh, that’s supposed to be a dystopia best avoided. (Ironically, my Secret Lab Magnus Pro desk has a Cyberpunk 2077 theme lol.)