Post B4pRf7f7yyP148wroW by reiddragon@fedi.catto.garden
(DIR) More posts by reiddragon@fedi.catto.garden
(DIR) Post #B4okI3KzbF08xnHdc8 by khm@hj.9fs.net
1 likes, 1 repeats
https://cvsweb.openbsd.org/log/src/usr.bin/tmux/screen-write.c,v?sort=Fileclaude output in baseopenbsd had a good run
(DIR) Post #B4opVEBv3aNWA28onw by pixx@merveilles.town
0 likes, 0 repeats
@khmDidn't they explicitly reject using it?Was this overlooked beforehand maybe?[Maintains forceful optimism]
(DIR) Post #B4opVEOKJSIqmVwj8y by khm@hj.9fs.net
1 likes, 1 repeats
no, of the big three only NetBSD has done the right thingthe person who committed that is the lead dev of tmux, which has a couple hundred words about which LLMs are Good, Actually, based entirely on whether their owners disclaim copyright on the output. A very confused understanding of how any of this works, with some bad priorities mixed in
(DIR) Post #B4ovXbNUQ6CgNR4Ipk by moody@hj.9fs.net
1 likes, 1 repeats
The change in question[0] was even made by an employee of Anthropic.[0] https://github.com/tmux/tmux/pull/4744
(DIR) Post #B4pD8UE9fPGo8RsMsK by asie@mk.asie.pl
0 likes, 0 repeats
@pixx@merveilles.town @khm@hj.9fs.net correct, but from the same effective position as the FSF: that LLM code is concerning from a copyright enforcement/liability perspective and no other reasonhttps://lwn.net/SubscriberLink/1064541/1a399d572a046fb9/
(DIR) Post #B4pEC3NxVRSq00j61I by mirabilos@toot.mirbsd.org
0 likes, 0 repeats
@moody @khm /cc @jessebot for addition to https://codeberg.org/small-hack/open-slopware (via https://polymaths.social/@rl_dane/statuses/01KM3J69F5BGJ6M5MC00N80EQ0 by @rl_dane ) for both tmux itself and, sadly enough, OpenBSD
(DIR) Post #B4pEKrEc3GPav8YLFw by mirabilos@toot.mirbsd.org
0 likes, 0 repeats
@khm @pixx /cc @jessebot for more on tmux
(DIR) Post #B4pEaAh7yy6HcOC0xM by anthk@neopaquita.es
0 likes, 0 repeats
@khm Tmux being either forked or dropped with abduco+dvtm joining base (ISC/MIT).https://github.com/martanne/abduco
(DIR) Post #B4pEaAspHTSSCffMBs by mirabilos@toot.mirbsd.org
1 likes, 0 repeats
@anthk @khm I’ll personally stick with GNU screen, but it’s good to know that I can put “look at whether tmux is good enough meanwhile, learn it, and add the missing features if still missing, then switch to it” from my TODO 😿
(DIR) Post #B4pEe6uhyvwb3AQJ04 by mirabilos@toot.mirbsd.org
0 likes, 0 repeats
@anthk @khm well, not good good to know, but YKWIM. Bit sad that the copyleft side seems to hold out slightly more…
(DIR) Post #B4pFa1wec4D4R5x45Y by prahou@merveilles.town
0 likes, 0 repeats
@khm chat, is it over?
(DIR) Post #B4pFa26vzqQuwylH72 by mirabilos@toot.mirbsd.org
0 likes, 0 repeats
@prahou @khm it’s not over yet #RunBSD #fuckAI
(DIR) Post #B4pFd5OAGXW3z9HCi0 by khm@hj.9fs.net
0 likes, 0 repeats
Short answer: no, it’s not “over” — but something did end.Here’s the real situation, without the hype:🧠 What actually happenedProjects like OpenCode were using Claude in ways they weren’t supposed to—specifically, routing consumer subscriptions (Claude Code) into third-party tools.Anthropic cracked down hard:Blocked spoofing of the internal Claude Code clientTightened token validationThreatened or took legal action against violators👉 Translation: the “cheap unlimited backend via subscription” trick is gone.🧩 About the OpenBSD / Claude commit thingIf you’re referring to OpenBSD commits mentioning Claude as a co-author or contributor, that’s mostly:Developers experimenting with AI-assisted patchesGit commit metadata including “Co-authored-by: Claude”That’s symbolic, not existential. It doesn’t mean:OpenBSD is “run by AI”Or human maintainers are goneIt’s closer to using a smarter autocomplete.🔥 So… what is over?These things are over (or mostly):Using Claude subscriptions as a cheap API backend in tools like OpenCodeThe “free lunch” era of agent swarms running on flat-rate plans🧭 What is NOT overOpen-source projects like OpenBSDAI-assisted codingThird-party tooling (they’ll just use proper APIs now)In fact, this pushes things toward:Paid APIsLocal modelsHybrid setups🧠 Reality checkIf your gut reaction is “it’s over,” that’s mostly the vibe from Twitter/HN drama.The actual shift is more boring:The loophole closed. The ecosystem adapts.If you want, I can break down the OpenBSD commit specifically (what was committed, how much was AI-generated, and why people freaked out).
(DIR) Post #B4pFnvWvGDoQauXgem by reiddragon@fedi.catto.garden
0 likes, 0 repeats
@khm link times out
(DIR) Post #B4pFnvhubMbR8zgSmm by cinap_lenrek@hj.9fs.net
0 likes, 0 repeats
https://hj.9fs.net/moody/p/1774970470.469575CC: @khm@hj.9fs.net
(DIR) Post #B4pFudlVhwf3WcZD9M by rl_dane@polymaths.social
0 likes, 0 repeats
@mirabilos @moody @khm @jessebotDang, what? #OpenBSD allows slop now? How open are the gates?
(DIR) Post #B4pFudxuxoaO96N7UO by mirabilos@toot.mirbsd.org
1 likes, 0 repeats
@rl_dane @moody @khm @jessebot tl;dr tmux does, and apparently the tmux lead dev is pro-slop, and someone (the cvsweb is down, cannot easily check) committed the slop unfiltered straight to OpenBSD before even morging it to tmux on Microslop GitHub. The slop was contributed by an Anthropic employee, even.I’m with “if Theo reverts this” but add “and revokes the commit bit of the person committing it” for OpenBSD, but tmux is lost.
(DIR) Post #B4pG2Mgn52V0sh2hnM by mirabilos@toot.mirbsd.org
0 likes, 0 repeats
@rl_dane @moody @khm @jessebot ah, it loaded. Committer was nicm@ and even mentioned the slop being slop in the commit message.Theo, the world is now looking at you.
(DIR) Post #B4pGDicBtv7m4IIyMi by mirabilos@toot.mirbsd.org
0 likes, 0 repeats
@rl_dane @moody @khm @jessebot I mean, tmux has several red flags already, use of Microslop GitHub, use of Googleslop Groups, …(nicm is also the tmux lead)
(DIR) Post #B4pGiCgSfbXs0qq4Q4 by KatS@chaosfem.tw
0 likes, 0 repeats
@khm @prahou A breakdown would be good; yes, please.
(DIR) Post #B4pGiD11R9zZ2cSUT2 by khm@hj.9fs.net
0 likes, 0 repeats
someone from anthropic submitted support for arcane shit to tmux, and the tmux dev committed it. since tmux's source tree is part of openbsd, there's now claude output in openbsd.so far it's the only verifiable slop in base.CC: @prahou@merveilles.town
(DIR) Post #B4pHokE2yY5ljSEOB6 by KatS@chaosfem.tw
0 likes, 0 repeats
@khm @prahou Almost secondhand contamination, then. Not great, but ould be worse.
(DIR) Post #B4pHokPOINAMIdXRrM by khm@hj.9fs.net
1 likes, 0 repeats
good summary, yeah. it happened in December and there's been no noise or pushback visible from the outside, so I find it low-key worrying nonethelessCC: @prahou@merveilles.town
(DIR) Post #B4pI00QF64aR9PIz5s by prahou@merveilles.town
0 likes, 0 repeats
@khm this is fine art.
(DIR) Post #B4pI00dMJJ4vo5RSXQ by khm@hj.9fs.net
0 likes, 0 repeats
I did my best to slightly misunderstand the question and then ran it through an "AI detector" and changed it until I got the score as high as I could. then I formatted it in a manner that would be appreciated by an asshole
(DIR) Post #B4pI93wZNEmWg7kDrs by reiddragon@fedi.catto.garden
0 likes, 0 repeats
@cinap_lenrek @khm saw these posts but I wanted to see the link in the original post
(DIR) Post #B4pI946ql10NC0YQtM by khm@hj.9fs.net
0 likes, 0 repeats
CC: @cinap_lenrek@hj.9fs.net
(DIR) Post #B4pJ3C2aSLLm0XKAOO by prahou@merveilles.town
0 likes, 0 repeats
@khm right back atcha!it only works as fine art as long as it's obvious that it's a post made by you though.
(DIR) Post #B4pJ3CFhfZqGfDSdpw by khm@hj.9fs.net
0 likes, 0 repeats
woops. should have been "a manner that would only be appreciated by an asshole"it was me all along
(DIR) Post #B4pKzKDmIbm083qGsS by crowbriarhexe@tech.lgbt
0 likes, 0 repeats
@khm @prahou you totally got me and I’m glad I kept reading the replies, otherwise I’d have angrily written you off as a slop apologist 😂
(DIR) Post #B4pKzKg8bASTY16v56 by khm@hj.9fs.net
0 likes, 0 repeats
four people blocked me after I posted that, presumably either because they thought it was pro-slop or because they thought it was anti-slop, so I must conclude I am surfing that knife edgeCC: @prahou@merveilles.town
(DIR) Post #B4pMcFya7G0mD8rV1E by mirabilos@toot.mirbsd.org
0 likes, 0 repeats
@khm another thing they might wish to be aware is that “XLibre” is fashtech… https://marc.info/?t=177477462000001&r=1&w=2
(DIR) Post #B4pNc206B0waox1n7Y by mirabilos@toot.mirbsd.org
0 likes, 0 repeats
@rl_dane @moody @khm did anyone report this to the devs? misc@ and tech@ have nothing this month yet…
(DIR) Post #B4pO1nGzYWMmfoDKfg by prahou@merveilles.town
1 likes, 0 repeats
@khm @crowbriarhexe you're an artist now, child
(DIR) Post #B4pO1nZmQfOZc50KxM by khm@hj.9fs.net
0 likes, 0 repeats
is there medicine I can take or
(DIR) Post #B4pO9ZDRhhvTOOGJ3w by rl_dane@polymaths.social
0 likes, 0 repeats
@mirabilos @moody @khmI'm not wading into those waters. XDI would consider you competent and able to do so. :D
(DIR) Post #B4pO9ZQutchY4AZ43k by mirabilos@toot.mirbsd.org
0 likes, 0 repeats
@rl_dane @moody @khm it’s… good if my name is not mentioned on OpenBSD mailing lists.I dared to fork it.
(DIR) Post #B4pOWOuGCTwS4wAykS by jessebot@social.smallhack.org
0 likes, 0 repeats
@mirabilos @rl_dane @moody @khm hoi, please use the repo for all communication on this project :heart_cyber:
(DIR) Post #B4pOWPNKSPC5X5mC3c by mirabilos@toot.mirbsd.org
0 likes, 0 repeats
@jessebot @rl_dane @moody @khm understandable, I don’t have the spoons for that atm but won’t bother you here.
(DIR) Post #B4pPKRWoj5wCgo4ZVo by rl_dane@polymaths.social
1 likes, 0 repeats
@mirabilos @moody @khmOh. I thought that would give you clout. :DI'm a know-nothing nobody to them, and I'm ok keeping it that way. XDkhm, how are your asbestos vestments? XD
(DIR) Post #B4pPKRqfXHojgNMQSG by mirabilos@toot.mirbsd.org
0 likes, 0 repeats
@rl_dane @moody @khmOh. I thought that would give you clout. :DThey really dislike people forking off them, and they mostly dislike people pointing out that they’re a NetBSD fork. *shrug*I wish I hadn’t needed to fork.
(DIR) Post #B4pPZdkBaiLqv3ZY8G by rl_dane@polymaths.social
0 likes, 0 repeats
@mirabilos @moody @khm> They really dislike people forking off them, and they mostly dislike people pointing out that they’re a NetBSD fork. shruglol kinda ironic, but ok. ;)> I wish I hadn’t needed to fork.I'd be curious to hear/read that story someday. ^___^
(DIR) Post #B4pPZe5oIJeI07goq0 by mirabilos@toot.mirbsd.org
1 likes, 0 repeats
@rl_dane @moody @khmI wish I hadn’t needed to fork.I'd be curious to hear/read that story someday. ^___^I wanted stuff. I mostly got ignored. Theo didn’t want stuff. Theo explicitly didn’t want wtf. At that point, I labelled my patchkit basically a fork, soft-fork at first.
(DIR) Post #B4pPiENxF8GesSGoym by khm@hj.9fs.net
0 likes, 0 repeats
not really into going into someone's project space and bitching at themI expressed my disappointment in my own spacebesides, if they didn't care in december I wouldn't expect them to care nowCC: @mirabilos@toot.mirbsd.org @moody@hj.9fs.net
(DIR) Post #B4pPlvad4BTRk4xg1o by mirabilos@toot.mirbsd.org
0 likes, 0 repeats
@rl_dane @moody @khm ah, and, episode Ⅱ, I met Theo at FOSDEM where he got some price, quickly talked about nroff/ditroff, and handed him a BSD shirt (with all four that existed at the time of printing on the back) as thanks, and later someone found a video from EuroBSDcon where they (not Theo though, some of the Germans after probably too much beer) put up a paper sign thanking me and burnt the t-shirt.
(DIR) Post #B4pPpWAnVUJW4XahE0 by mirabilos@toot.mirbsd.org
0 likes, 0 repeats
@khm @rl_dane @moody no bitching required, just pointing out to them that one of their devs committed positively slop to their repo
(DIR) Post #B4pPtGWncJlOhxfXjE by mirabilos@toot.mirbsd.org
0 likes, 0 repeats
@khm @rl_dane @moody in case they didn’t know
(DIR) Post #B4pQWWvQaPhirBua2a by rl_dane@polymaths.social
0 likes, 0 repeats
@mirabilos @moody @khmlol what?!? Why??
(DIR) Post #B4pQWX8tmKTnWyDL2O by mirabilos@toot.mirbsd.org
0 likes, 0 repeats
@rl_dane @moody @khm *shrug* beer?
(DIR) Post #B4pQaBeFdTqO1ecZHc by rl_dane@polymaths.social
0 likes, 0 repeats
@mirabilos @moody @khmJust curious, why fork OpenBSD vs. working with NetBSD or maybe even forking it?
(DIR) Post #B4pQaBml7qeKS2bMXo by mirabilos@toot.mirbsd.org
0 likes, 0 repeats
@rl_dane @moody @khm because that’s what I use(d)… NetBSD was really old-smelling back then, and they don’t have basic things like lynx in base. And despite trying, they still haven’t moved their in-base pdksh to mksh (there are (minority and mostly dfbsd-aligned) voices indicating they’d rather drop all shells not sh to pkgsrc).(Incidentally, eliminating ash derivatives from the world is a lowkey goal of mine.)
(DIR) Post #B4pRLNHOD08Pi4Qcsa by cinap_lenrek@hj.9fs.net
0 likes, 0 repeats
tmux1.mp3CC: @reiddragon@fedi.catto.garden
(DIR) Post #B4pRf6jLRoBKAvuikS by rl_dane@polymaths.social
0 likes, 0 repeats
@khm @prahouYeah, that was gross. 🤣(Re: https://hj.9fs.net/khm/p/1774986433.595571)
(DIR) Post #B4pRf70iPE4n2o2ap6 by feld@friedcheese.us
0 likes, 0 repeats
@rl_dane @khm @prahou Any contributor in any project can be doing it without you knowing, so if you REALLY want this kind of purity the answer is to stop taking contributions from other people as you can't verify the provenance of the code.
(DIR) Post #B4pRf7DpcSZHhUB4Ge by jackdaniel@functional.cafe
0 likes, 0 repeats
@feld @khm @prahou @rl_dane the same could be said about any contributor being able to copy paste proprietary code; if we abandon assumption of honesty, then external contributions are no go -- LLM or not.
(DIR) Post #B4pRf7OoxbMIFZJqOe by feld@friedcheese.us
0 likes, 0 repeats
@jackdaniel @khm @prahou @rl_dane so where do we go from here? Because if people keep ramping up this rhetoric the only option is to abandon trust by default. Now you're a team developing in the open, but you're not openly accepting anything from people you don't know.I know a many people across a bunch of very high profile projects that are using LLMs in their workflows now and just not disclosing it because they don't want to deal with the drama. The code is passing the smell/review tests and getting merged because there's nothing wrong with it: it does what it's supposed to, it follows the project's existing standards and style guidelines.Good luck finding out which developers are doing it and banning them from the projects. :)Unless there's a legal precedent that the generated code is definitely a violation or now must have a viral license attached with it, nothing will change. I have very very strong doubts about both of those happening.Why?Because this is *our* legal system. These are our own chains we'd be putting on ourselves. And our governments will see this very differently. If our competitors in the global market will not be enforcing these laws on their citizens, why should we do it to ourselves? They'll be able to build faster than we can because they're not weighed down by a legal albatross that in many respects is just silly and incompatible with modern technology.
(DIR) Post #B4pRf7f7yyP148wroW by reiddragon@fedi.catto.garden
0 likes, 0 repeats
@feld @jackdaniel @khm @prahou @rl_dane > The code is passing the smell/review tests and getting merged because there's nothing wrong with itsure, if you ignore the ethical, social, and environmental issues along with the ties to fascists.In other words, you might as well be saying Trump's not all that bad if you just ignore how he wrecked the global economy and started another fucking war in the middle east and sent ICE to kidnap and kill people within the US.
(DIR) Post #B4pRf7vR0LRjsiZtEO by khm@hj.9fs.net
0 likes, 0 repeats
the person you're arguing with is an inveterate teslafucking bootlicker; they exist only to tell you you're wrong because whatever they like is morally justified thereby.just take a glance at their public posts: LLMs are fine but typing indicators in chat apps are unsustainable resource hogs. such dipshittery is the primary output there.nearly identical behavior, according to my notes, is why I blocked them approximately six thousand years agoCC: @jackdaniel@functional.cafe @prahou@merveilles.town @rl_dane@polymaths.social
(DIR) Post #B4pS6udxymFYnUeliS by jbowen@mast.hpc.social
0 likes, 0 repeats
@mirabilos @rl_dane @moody @khm @jessebot Time to go lurk in their IRC...
(DIR) Post #B4pS6usV6jsNWZSNN2 by mirabilos@toot.mirbsd.org
0 likes, 0 repeats
@jbowen @jessebot @khm @rl_dane @moody ICB… until they boot you
(DIR) Post #B4pSfaNJKUQ44ROpFY by rl_dane@polymaths.social
0 likes, 0 repeats
@mirabilos @jbowen @jessebot @khm @moodyResults for ICB [ÖBB] Intercitybus Internet Citizens’ Band Intracerebral bleeding (intrazerebrale Blutung)—http://www.mirbsd.org/wtf.cgi?q=icb?? ;)
(DIR) Post #B4pSfampnaptLbLD28 by mirabilos@toot.mirbsd.org
0 likes, 0 repeats
@jbowen @khm @rl_dane @moody the middle one ;)
(DIR) Post #B4pVpO5KB7PKiCLsfY by jackdaniel@functional.cafe
0 likes, 0 repeats
@feld @khm @prahou @rl_dane this is a tired line of argumentation. extensive LLM usage contributes to climate crisis and society crisis.Laws are for us humans, indeed, and if some activity brings more harm than good, then it should be regulated or outright banned.This is just a side note, but after people stop creating things on their own, models will rather quickly degrade, that's why llm companies are so desperate to suck in any "organic" data from the internet, even at the cost of scraping the whole internet over and over again.
(DIR) Post #B4pVpOgBy2tkYXbK8O by feld@friedcheese.us
0 likes, 0 repeats
@jackdaniel @khm @prahou @rl_dane > this is a tired line of argumentation. extensive LLM usage contributes to climate crisis and society crisis.this is fearmongering and a lie. LLMs don't *require* a crazy amount of energy or compute resources to operate; this is simply the current status because of very poorly optimized inference engines.Everyone in the industry knows it, but it's the thing you just don't say out loud because the industry relies on this circle jerk of money flowing between Nvidia, OpenAI, etc etc.What happens to OpenAI if you could host a thousand concurrent users on a single server hosted in your own company's datacenter?What good is it for Nvidia to follow through on their research papers pre-2020 discussing how to build these inference engines properly to reduce resource requirements if it's going to destroy your market cap because now nobody needs to order all these monster GPUs?Right now, the waste is intentional, but the new engines will evaporate all of this.
(DIR) Post #B4pVpP9GDy9O0hCXRY by sun@shitposter.world
0 likes, 0 repeats
@feld @jackdaniel @khm everything "contributes to the climate crisis" and what we choose to keep doing is a value judgment so this argument is just a facade for "I don't like it so stop doing it".
(DIR) Post #B4pbMfQpChalq564VE by eukara@idtech.space
0 likes, 0 repeats
@mirabilos @rl_dane @moody @khm pretty sure the ext4 thread should be notified https://marc.info/?l=openbsd-tech&m=177411552501319&w=2
(DIR) Post #B4pbMfmnszAmwFNclE by mirabilos@toot.mirbsd.org
0 likes, 0 repeats
@eukara @khm @moody @rl_dane nah, it’s a separate thing
(DIR) Post #B4pbuyesE5ejJb1pCa by rl_dane@polymaths.social
0 likes, 0 repeats
@mirabilos @jbowen @khm @moodyAhh, I'd never heard that name before, but I love it. :D
(DIR) Post #B4pbuyondBazoNfkfo by mirabilos@toot.mirbsd.org
0 likes, 0 repeats
@rl_dane @jbowen @khm @moody meh, I’ll stick with IRC
(DIR) Post #B4pe9nlBEBpTlbAtU0 by pernia@cum.salon
0 likes, 0 repeats
@sun @feld @jackdaniel @khm didnt google already release an AI thing that uses way less memory? it tanked a bunch of ram companies apparantly. it might have been bs though i didn't check the source i read it from.
(DIR) Post #B4pe9oBPeeoT4xRqN6 by sun@shitposter.world
0 likes, 0 repeats
@pernia @feld @jackdaniel @khm my assumption is that usage of AI will just grow to capacity so it won't ever really save water or electricity or compute
(DIR) Post #B4peLBgId5Pz8uJxtQ by pernia@cum.salon
1 likes, 0 repeats
@sun @feld @jackdaniel @khm they'll probably have to optimize after the energy crisis hits.also i always hear this shit about "water" but its just for industrial cooling right?
(DIR) Post #B4peX4udcZnfcnhK40 by feld@friedcheese.us
1 likes, 0 repeats
@pernia @jackdaniel @khm @sun and ignores that there are closed loop water systems used for cooling...
(DIR) Post #B4rAeRaV6ZlHfXnKdM by inata@types.pl
0 likes, 0 repeats
@khm hahahaha the patchset is a fix to make it so that Claude Code, a react app that renders to the terminal at 60fps, doesn't flicker when it rendere to the terminal at 60fpsit's so over
(DIR) Post #B4rAeRqS9GWQT1G4Uy by khm@hj.9fs.net
0 likes, 0 repeats
yeah the actual goals of the patchset are a completely separate clown showstupid plans executed stupidly
(DIR) Post #B4rCzgLHdvCdQ2vpce by inata@types.pl
0 likes, 0 repeats
@khm i feel like tmux has been less lenient with awful new terminal specifications in the past (hint: most of these terminal specifications are spearheaded by the fancy-new-terminal-emulator dev du jour) so i'm sad this is the one that does them in. oh well the terminal was a joke anyway back to unstructured unicode output for me
(DIR) Post #B4rCzghcIt4EXJNfQu by khm@hj.9fs.net
1 likes, 0 repeats
looking forward to ignoring the kitty TermGL shader protocol with 5.1 Audio
(DIR) Post #B4rzIvErHJiyvBAOQK by jens@social.finkhaeuser.de
0 likes, 0 repeats
@mirabilos @rl_dane I've got to ask... how you'd go about it?Context: been installing it again for the first time since the early 2000s, and am have been positively surprised.Now I am wondering how to make changes that are less "add a package" and more "maintain an additional system layer made up of a bunch of packages". It seems as if it was sort of made for this?But I struggle to find an entrypoint.If you "forked it", it seems likely you'd have some pointers?
(DIR) Post #B4rzIvTkNxdNfM8HdA by mirabilos@toot.mirbsd.org
0 likes, 0 repeats
@jens @rl_dane I didn’t do much with packages at first (save for one extra). I had changes in base, so to src/ not ports/.First I kept a patchkit (as Ms in the working copy). Later, with the wtf rejection, I started a CVS repo and imported a checkout of OpenBSD into it, checked that out, applied the patches I already had, and continued development from there.So, not very layer.If you want to keep a bunch of local packages in a tight repo, I’d make /usr/ports/mine a repo (can be RCS, CVS, git, whatever) and put your packages in there. That way you can version them and they will build and install just fine. Make sure to not commit the w-* directories, that is, only commit in a make clean state.Does this help? (If not, maybe explain more what you have planned.)
(DIR) Post #B4sD1qj8CME5Mc50cK by jens@social.finkhaeuser.de
0 likes, 0 repeats
@mirabilos It helps a lot, thanks!My first goal is to have a bunch of packages consistently there across multiple machines, so making "mine" a repo is certainly the thing that'd get me there.Then it's was speculating: what else would it take to have that collection not be ports, but just another tarball you can select during installation?But that part isn't actually all that important if I'm being honest.I guess the thinking there goes in the direction of having an...@rl_dane
(DIR) Post #B4sD1r0V9m7YEUCsgy by mirabilos@toot.mirbsd.org
0 likes, 0 repeats
@jens @rl_dane these tarballs are called 'sets'.I’d not go that route, because the sets are (naturally) outside of packages, and so they cannot be upgraded, etc.What you can do, if the machines are identical in CPU architecture and OS release, is to have one build binary packages (if they haven’t changed it, this happens anyway if you install them, look in /usr/ports/Packages/) and expose them via HTTP, then just pkg_add http://… to install them on the others, which is easily scripted.If you really want to go the set route, you basically go into compile-on-your-own land. For something with autoconf, the route would be, roughly:(cd /usr/src && make distrib-dirs DESTDIR=/tmp/foo)cd /tmp/foosudo find * | sort >/tmp/x1cd -sh configure --prefix=/usr --sysconfdir=/etc …makesudo make install DESTDIR=/tmp/foocd /tmp/foosudo find * | sort >/tmp/x2comm -13 /tmp/x{1,2} | sudo cpio -oC512 -Hustar | gzip -n9 >/path/to/setnameWhere setname would be something like site35.tgz (“site” is, I believe, recognised as præfix).This of course makes a lot of assumptions, like that the thing does honour DESTDIR on install, and you review the list of files/directories it adds (comm -13 /tmp/x{1,2}) to make sure it’s good first, etc.
(DIR) Post #B4sD1rEgJ3SmwSqCnI by jens@social.finkhaeuser.de
0 likes, 0 repeats
@mirabilos... alternative to the pretty hacky embedded Linux things out there. Ignoring what hardware support exists or does not for a moment, you sort of want two things: be able to build images (works easily somehow, haven't tried it out), and basically build those as "base system+feature set(s)"For example, I'm thinking of the robot I'm tinkering with. Would be very nice to have (no idea if compatible), base + ros2 [+ my robot packages] as an easy selection.Since install...@rl_dane
(DIR) Post #B4sD1riSWLHaQolzCy by jens@social.finkhaeuser.de
0 likes, 0 repeats
@mirabilos... allows you to select e.g. X related stuff like that, I was wondering how that worked, how to add something else, etc.If "mine" is a way to get me halfway there, it's likely good enough.@rl_dane
(DIR) Post #B4sDDKiN2qfxYRCSHY by mirabilos@toot.mirbsd.org
0 likes, 0 repeats
@jens @rl_dane you absolutely do not want to build images. That will clone things that should be individual and random, from the seed file over the ssh keys to the inode generation numbers which are relevant for at least NFS.Best to make a way to make it easy enough to install multiple.If you want arbitrarily-named sets, you might have to patch around in /usr/src/distrib/common/ and build your own installer (for which you’ll have to build most of base (all of it is easier but takes longer), but that should not be a problem).
(DIR) Post #B4sDSes8NkxWPp4fOS by mirabilos@toot.mirbsd.org
0 likes, 0 repeats
@jens @rl_dane all of this will still be easier and more compartmentalised than doing a full softfork.Mind you, use a release before Dec 2025 to avoid the slop.
(DIR) Post #B4sDm9VwgCZox6zW9w by mirabilos@toot.mirbsd.org
0 likes, 0 repeats
@jens @rl_dane in favour of the “/usr/ports/mine/” way:You can do metapackages that depend on a list of others from both your set and the normal packages, and then you have only one package to pkg_add and it’ll acquire the rest automatically.
(DIR) Post #B4sEFrnh6TG4hZShYe by mirabilos@toot.mirbsd.org
0 likes, 0 repeats
@jens @rl_dane I have been pondering moving parts of base to ports, to make them upgradable more easily, as “base packages” that end up in /usr instead of the ports præfix, and some machinery to bundle them up into a set, but never had enough spoons to do that… if you already build the full OS, base is easier to work with, but you lose all the benefits from using a package system, like upgrades, partial uninstallation, etc.
(DIR) Post #B4sKB5Jo13hIu6aJxg by jens@social.finkhaeuser.de
0 likes, 0 repeats
@mirabilos @rl_dane Fair enough, and this makes a lot of sense!There are a lot of issues I have with the specific choices made, but I generally like the idea of layering my system. In a sense, this is not good for upgrade reasons. But it's hard to manage well othwrwise.Metapackages really are the choice here. I wasn't sure they existed here, but if they do, that's perfect. Then I don't really need sets, no.I guess that's the only thing I find a bit difficult to get used to with BSDs, ...
(DIR) Post #B4sKB5XHCyTNZst4xU by jens@social.finkhaeuser.de
0 likes, 0 repeats
@mirabilos... that split between base, ports and packages.But it's far from impossible!So practical management question, do you fork the ports tree to maintain changes there? I mean, there's a comprehensive guide on that I could read... no need to reply! I just hadn't gotten around to that.@rl_dane
(DIR) Post #B4sKB5jgSqOiCMgzIW by mirabilos@toot.mirbsd.org
0 likes, 0 repeats
@jens @rl_dane in my case, I eventually hard-forked the OS. That usually ends in inability to keep up with changes. In my case, the interest also was not ports.If you just want to maintain some local packages, new or changed, then putting them into a subdir that you version separately is the way to go.
(DIR) Post #B4tI973Eb6Cgwrj33I by rl_dane@polymaths.social
0 likes, 0 repeats
@jens @mirabilosI don't know if this is helpful, but this is how I think about it (coming from being a total n00b):base: think "immutable linux," but without the containerization buzzword magic. This is the software that comprises the OS itself. Sadly, the linux world doesn't quite grasp the concept of an OS. Not at that level.ports: think AUR/yay/yaourt/etc. A superset of the packages (I believe), compiled from source. Generally just a huge directory structure of Makefiles.packages: think pacman
(DIR) Post #B4tI97k81cVz5tnIuW by jens@social.finkhaeuser.de
0 likes, 0 repeats
@rl_dane @mirabilos Misunderstood the question, I think?I get the difference. I don't get (other than for historical reasons) WHY this difference.In the sense that one might want to manage the OS and packages a bit separately? Yes. Absolutely. But also, the base+patches thing, and the ports/packages thing are two entirely orthogonal systems which solve more or less the same overall issue, which is define the packages to be installed, and allow for upgrades.If ports have metapackages...
(DIR) Post #B4tI98RNQp6rG21qK0 by mirabilos@toot.mirbsd.org
0 likes, 0 repeats
@rl_dane @jens base is base, it’s the OS, period. base is worked on as a whole, no separation into packages like in the GNU world.Ports are just a mechanism to add third-party stuff to your installtion for convenience.
(DIR) Post #B4tI98pTzCOMSnJ5tY by jens@social.finkhaeuser.de
0 likes, 0 repeats
@rl_dane @mirabilos ... an OS set is like a metapackage, a patch like an update to a package or metapackage.Seems weird.Mind you, that's not to suggest I want to throw out the baby with the bathwater and wish for hundreds of individually installed packages for my base OS. It works for Debian, beyond that, a lot of distros seem to struggle with this more.Hence this notion of a "layer". Something like a metapackage you can apply to a base OS to get some functionality.Sets are the same...
(DIR) Post #B4tI99STeDaGPjYEfw by jens@social.finkhaeuser.de
0 likes, 0 repeats
@rl_dane @mirabilos ... concept, in the abstract.Anyway... it's mostly idle thoughts.
(DIR) Post #B4tIEcqzmdcl1wd6SO by mirabilos@toot.mirbsd.org
0 likes, 0 repeats
@rl_dane @jens sets are just a way to not install all of base, for size reasons (when hard discs were sized in Megabytes)