ninjin

There are some library functions to support databases in base, but that is probably not what you are asking for. Likewise, to the best of my knowledge there is no TUN/STUN server. However, you likely have both in ports.

Not sure if you are asking these as "Gotcha questions", if not, sorry, reading tone, etc. over the Internet is hard. If you are though, you are missing the point of what OpenBSD is and asking those questions is akin to walking into a Burger King and asking for sushi or how Linux interoperates with proprietary Windows drivers.

OpenBSD is the answer to what happens when a bunch of programmers get together around a shared history (BSD), goals (security, sane defaults, etc.), and very limited resources (Linux may be on the order of magnitude of a million times more users and funds). For example, based on reading tech@ for a few years, I am certain that the OpenBSD developers would love to have a new file system (people on Hacker News love to complain about FFS2), but the problem is that they are acutely aware of their own limitations. Ignoring the licensing issues with say ZFS, they do not have the funds and manpower to bring it into base (and maintain it as well, which everyone conveniently seem to forget is a cost) given how complex and large of a piece of code it is. So, an OpenBSD solution to what a new file system will look like will always be different from what we see out of Windows, Linux, etc. This, to me, is a good thing, because I like software diversity and history has shown that out of the OpenBSD community comes amazing pieces of software that makes its way into the rest of the ecosystem: OpenSSH, OpenSMTPD, tmux, LibreSSL, etc.

As a user, it forces you to rethink the cost of what you run and to some degree change your habits. I, for example, thought I "needed" a media server, but instead I have a directory serving videos over HTTP via httpd(8) and I just use the default directory listing and copy a URL into VLC to watch. Now, you may cry "That is not the same as a media server!" and I am not claiming it is, but I am getting along with no dependencies and fewer lines of code and am happily watching my videos regardless of any objections. Plus, I bet my "media server" upgrades come with a lot less drama.

> Is it even valid to have additional restriction on top of Apache 2.0?

You can legally do whatever you want, the question is whether you will then for your own benefit be appropriating a term like open source (like Facebook) if you add restrictions not in line with how the term is traditionally used or if you are actually be honest about it and call it something like "weights available".

In the case of OpenAI here, I am not a lawyer, and I am also not sure if the gpt-oss usage policy runs afoul of open source as a term. They did not bother linking the policy from the announcement, which was odd, but here it is:

https://huggingface.co/openai/gpt-oss-120b/blob/main/USAGE_P...

Compared to the wall of text that Facebook throws at you, let me post it here as it is rather short: "We aim for our tools to be used safely, responsibly, and democratically, while maximizing your control over how you use them. By using OpenAI gpt-oss-120b, you agree to comply with all applicable law."

I suspect this sentence still is too much to add and may invalidate the Open Source Initiative (OSI) definition, but at this point I would want to ask a lawyer and preferably one from OSI. Regardless, credit to OpenAI for moving the status quo in the right direction as the only further step we really can take is to remove the usage policy entirely (as is the standard for open source software anyway).

Official website with links to Bandcamp, live streams, IRC, tour schedule, etc.

https://mbrserver.com

Also, DRM-free music and streaming for KEYGEN CHURCH on Bandcamp:

https://keygenchurch.bandcamp.com

I have had success running a hybrid IPv4/6 network by reading this guide for inspiration:

https://blog.infected.systems/posts/2024-12-07-building-an-i...

This allows me to have a mixture of both protocols and even some boxes that have both IPv4 and IPv6 addresses. I still have some issues writing routing rules that does not fail for link-local addresses, but the network has now been fully operational for well over a month.

Link to the published paper rather than the preprint (update link?):

https://aclanthology.org/2025.acl-long.1126

That is not the same as in base though. For example, the NixOS developers maintain a large chunk code to generate scaffolding around systemd and to build code via Nix. They do not take a web server, fork it, and maintain it (alternatively, write that web server from scratch as is the case for httpd(8)). When I set up that OpenBSD server, I install the base system, place the configurations, start the daemons, and I am set. Not a single line of code runs outside of what is in the base repository.

I realise that the way BSDs do things is very different from Linux, but in BSD land the same people write the kernel, user land, and maintain the ports tree. With this I am not saying it is superior, but it does lead to a very different experience both as a developer and user. Yes, there are some exceptions to this like clang, the AMD GPU driver, etc. But the overall picture is true.

No worries, glad to help. I should also have added that on OpenBSD (and BSDs in general) base is much more complete than on most Linux distributions. I run a web, name, and mail server using OpenBSD and I do not need a single port or package for this. SMTP daemon, DNS daemon, web server, TLS certificate handling, etc. It is all in base and works together coherently, which is what draws people like me to BSDs.

It is not an excuse, you are perhaps just not familiar with how software is handled by OpenBSD and other free operating systems. You have a base system, which includes the kernel, user land, etc. Then on top of this you have ports and packages, usually in a different source tree.

Unless you have a nearly insane amount of resources, vetting all those ports and packages is incredibly hard if not impossible (think Canonical and Red Hat level of funding, when on a good year OpenBSD raises ~USD 500,000 and Red Hat's revenue is in the billions) and even then there will generally be different tiers of support. Thus, the strongest guarantees will always be with the base system as it is installed by everyone and outside of Linux also developed by a single team. Yes, plenty of ports such as Firefox on OpenBSD have some great security patches applied, but expecting all the over 10,000 ports to have the same level of quality and attention applied as what is in base is unrealistic no matter how much security and correctness is a priority.

I have maintained ports and packages across the Linux and BSD ecosystems for a good while now and I have more confidence in what OpenBSD has in ports based on my own experience compared to several Linux package managers. However, that does not mean that I will try to pull out a random OpenBSD package and expose it to the Internet before doing due diligence.

> I'm surprised to see this; my first thought was also something like

>> it's old unsecure code

> and that seems very unlike OpenBSD.

It is going into ports and I am positive there are plenty of worse offenders in terms of insecure software in there already. If this was going into base, I am confident the clean up would be substantial, but there is already cwm[1] and fvwm [2] for your modern and classic window manager needs in base.

[1]: https://en.wikipedia.org/wiki/Cwm_%28window_manager%29

[2]: https://en.wikipedia.org/wiki/FVWM

If you read the interview and walk away with that impression I will be amazed. You may not like his distinction between science and engineering, but he admits it is somewhat arbitrary and as someone that is solidly in the deep learning camp his criticism is not entirely unfair, even if I disagree with it and will not change my own course.

Personally, I find it somewhat amazing that you put Demis on that list given that he, himself, on very good accounts that I have, explicitly pushed back against natural language processing (and thus large language model) development at DeepMind for the longest of times and they had to play major catch up once it became obvious that their primarily reinforcement learning-oriented and "foundational" approaches were not showing as much promise as what OpenAI and Facebook were producing. Do not get me wrong, what he has accomplished is utterly amazing, but he certainly is not a father of large language models.

As someone having worked in natural language processing for nearly twenty years, I understand where you are coming from with this jab at Chomsky, but it is ultimately a misrepresentation of his work and position. Chomsky has to the best of my knowledge never showed any interest in building intelligent machines as he does not view this as a science. Here is a fairly recent interview (2023) with him where he outlines his position well [1]. I should also note that I am saying this as someone that spent the first half of their career constantly defending their choice of statistical and then deep learning approaches from objections from people who were (are?) very sympathetic to Chomsky's views.

[1]: https://chomsky.info/20230503-2

Cheers! They look like decent chaps and also outside the US for some additional certificate diversity. Are there other trustworthy Acme issuers out there?

A pity that acme-client(1) does not allow for fallbacks, but I will add a mental note about it being an easy enough patch to contribute if I ever find the time.

> I love OpenBSD and I'm really interested in got for several years. I just hope that this https://framagit.org/stsp/got/-/blob/main/got/got.c?ref_type... gets removed at some point.

It is an OpenBSD culture reference though [1], not some sort of attempt at an insult. I would personally be sad to see it go as it does make me smile when a program is not too "corporate" and reminds me every time I see it that I have the power to pull through on my own (most of the time) if I just put my mind to it.

[1]: https://www.openbsd.org/lyrics.html#51b

> Too bad managing [Wireguard] is such a pain[.]

It is? My clients on OpenBSD need about ten lines in /etc/hostname.wg[0-9]+ and that includes the routes!

But why fall back on the state to police every website and service? Not to mention that the Internet is wider than any jurisdiction you can hope to achieve.

Plenty of devices have parental controls and it is your duty as a parent to look into this and understand the consequences when you hand a device to your child that allows them to consume a global diet of media and connects them to a majority of the global populous. It is not unreasonable to expect device/software manufacturers to provide the power to parents to do this and we already have had the free market provide options for this for decades.

To me, it is not any different than to expect a parent to consider the potential consequences of children playing in the streets, etc. and take action. Thus, to me, the people pushing this kind of legislation are either sloppy thinkers or dishonest in what aims they hope to achieve.

> I don't know if it is the best, but the music of S3&K absolutely slaps. I have yet to play Thunder Force 4 though.

It is... Okay?

https://archive.org/details/sonic3_knuckles_ost

But just listen to say Metal Squad or Stand Up Against Myself from here:

https://archive.org/details/md_music_thunder_force_iv

More favourites of mine as I love the amount of character the Yamaha chip has:

https://archive.org/details/md_music_the_adventures_of_batma...

https://archive.org/details/md_music_battletoads

https://archive.org/details/StreetsOfRageMD

https://archive.org/details/castlevania-bloodlines_ost-1993

> Which Sonic game? The first one is quite bad, but S3&K is probably my favourite platformer ever.

Sonic the Hedgehog 1 and 2, plus Sonic Mania as those are the ones I have touched somewhat recently. I can not remember if I played Sonic 3 & Knuckles as a kid or not, so maybe I am overly biased. In general, I have a fairly positive view of the Sonic platformers in every regard apart from the controls. The music for example is good, although it is not the best that the Mega Drive has to offer as that would probably be Thunder Force IV. Same for having very nice sprite work and the Sonic 2 bonus level blew my mind as a kid.

> Regarding my comment about the difficulty, for me it is mostly a result of the bad controls, which to be honest is a problem that plagues many games of that era. You have enemies popping out from the edge of the screen and you need to react in a fraction of a second, but the controls don't let you do so, so you need to memorize the whole game and react in advance. It is not fun for me.

Can relate, but I wonder if what you are describing here is the excessively zoomed in trend that we had from the late 80s and throughout the 90s.

There is this whole "It is not a proper meal unless there is a bowl of rice" kind of thinking. So you get rice with the potatoes that come with your Nikujaga, the mashed potatoes, steak and fried potatoes, etc. But then there is that completely arbitrary line that you do not get a bowl of rice with your noodles (unless it is for dunking into the leftover soup of your Ramen or Tsukemen), pasta, burger, etc.

I love Japanese food culture to bits, but figuring out when a bowl of rice is vital for a meal to even be considered a meal and when it is not is still a mystery to me after observing my wife and in-laws for more than a decade. Having been taught the "food pyramid" [1] line of thinking as a child, it just makes no sense to this gaijin how we can not substitute one source of carbohydrates with another, but I suppose that is what culture is for you.

[1]: https://en.wikipedia.org/wiki/Food_Guide_Pyramid

Played a tiny bit of the first one in the 90s, but like you I only played through the trilogy just recently (in my case, last year) and on original hardware. I would not go so far as to say disliked, but you are absolutely correct that the controls are floaty (although not as bad as Sonic, but Sonic is truly awful) and it is often a challenge to tell where objects and platforms end. The third game has the weakest music, but I would consider a lot of the level and boss design substantially better. Overall, I think there is a heavy amount of nostalgia going on when it comes to the fans, but that is okay as long as we speak openly about it.

As for difficulty, we will have to agree to disagree. Only point that had me somewhat frustrated was the waterfall boss in the last game and that one had me stumped for some time (should have read the manual I guess?). Overall, I would still recommend the games as good for their time and among the better action platformers for their console, but they really are nowhere close to a masterpiece like Super Mario World that has pretty much perfected controls and you can tell exactly where platforms and objects start and end.

I fail to see how that is different. Ultimately, you have released a piece of software into the wild with a clause stating: "The software is provided 'as is' and the author disclaims all warranties with regard to this software including all implied warranties of merchantability and fitness". Thus, it is purely cultural that somehow others and yourself expect you to cancel your family time on a Saturday night solely because an issue has been found in a piece of software you have given away for free. This "value add" is wearing people out and if we want this expectation to remain, maybe it is time for those profiting or those with a monopoly on violence to explore ways to support those that kindly provide free labour like this?

It is reasonable to assume that this is true. But an equally effective way other than making your license unpalatable to them, is just to say no and state clearly: "Patches or GTFO". Also, have a homepage to link with your (hefty?) consulting rates?

I have mentioned this in the past, but there was this weird shift in culture just after 2000 where increasingly open source projects were made to please their users, whether they were corporate or not, and "your project is your CV" became how their maintainers would view their projects. It does not have to be this way and we should (like it seems to be the case with libxml2) maybe try to fix this culture?

"The reason why we have Linux, and BSDs, and XNU is that they all provide the same baseline API, which was defined from the outside [by POSIX]. The coordination problem was pre-solved, and what remained is just filling-in the implementation."

But that is not at all how Posix operates or has operated. Posix standardises common denominators between existing implementations. The fact that we now have strlcpy(3) and strlcat(3) in Posix 2024, is not because Posix designed and stipulated them. Rather, they appeared in OpenBSD in 1998, were found useful by other *nix-es over time, spread, and were finally taken aboard by Posix that standardised what was already out there and being used! This to me is the very opposite of the point the author is trying to make!

If the world is heading the way the US is heading, I may be inclined to start doing the same. This is the only "social media" account I have left, but if my freedom of expressing myself will be impaired by governmental stalking like this I will sadly have to "adapt". Losing my ability to help filter and manage by upvoting and flagging blows though.

I just do not understand the hostility where all there is is different approaches. A charitable take (there are plenty) for example would be that I (like many) have about twenty years of legacy gear investment that I need to consider and/or was unaware of the rather recently introduced (2018?) Nikon Z system. You clearly have "a system" that works for you and that is great! Can you even get autofocus on a 1930s Elmar with that setup? Because would just be bonkers to think of from a technical perceptive.

Personally, I prefer to shoot manual to keep the operation the same between film and digital and I am fast and competent enough with a rangefinder patch that I do not feel that autofocus gives me that much and I only shoot primes (28 and 50mm). The Nikon Z system looks really interesting though, so thanks for bringing it to my attention. It is great to see innovation in this space that is not just Fujifilm; I will make sure to try my hands on one next time I am in one of the big camera stores.

I think it is fair to mention the aspect of Leicas being fashion accessories, but it is equally unfair to pretend that they are just fashion accessories. Let me try to explain.

If what you are looking for is image quality (especially when considering quality per pound spent) Leica is not where you should be looking. However, they fill a very unique niche: small(-ish) and light(-ish), full-frame sensor, and operation that is nearly identical to a film rangefinder. What does this mean? It means that you can have a system where you shoot the same lenses, filters, etc. on both digital and film bodies and the cameras and lenses will behave the same way. This, in addition to me subjectively liking the way that rangefinders operate, is why I have a digital Leica. Although I unashamedly shoot Voigtländer and Zeiss film bodies (both made by Cosina in Japan) as they are compatible and come at a fraction of the cost of a film Leica. Before I head out, I ask myself: "Do I feel like film or digital today?" and then I pick the body, my favourite 50mm lens (because to me, that lens is so important that it dictates the rest of my system), and off I go.

You are correct that mirrorless cameras in many ways is where "the game" is right now, but do not make the mistake to counter the foolish Leica elitism with an equally foolish elitism of your own. As a photographer your aim should be to have a setup that works for you, to realise your vision, and while my setup works for me and my digital/film setup, I readily recommend others that do not care about film to explore the mirrorless Fujifilm X-series as they are fun to shoot and price effective. If you want to explore film, the sensible thing is to just pick up an old point-and-shoot or SLR before you invest into a costly system as maybe you will not like film in the end?

As for the "Leica fashion" market, I am actually kind of thankful for it as a Leica shooter because it fills up the second-hand market so that I can get two or three generations old digital Leica cameras at less insane prices. Although what I really wish for is a cheaper competitor to Leica with a full-frame, M-mount, digital alternative so that I could dump Leica and still have my system work. For example, the Epson R-D1 from 2004 had better ergonomics than any digital Leica until the M10 was released in 2017, but 6.1 megapixels and a sensor that becomes borderline unusable over ISO 400 is sadly not viable unless you are going for some sort of retro-digital look. So, the "sane" digital Leica choice these days is likely a used M10 or maybe M10-R (or a Typ 240 if you are "poor", like me) and the M11 largely looks like a lot of money for next to no benefit (but I am thankful that it exists and drives down the prices of the digital Leicas I want to shoot).

My favourite movie of all time:

https://en.wikipedia.org/wiki/Dr._Strangelove

The writing and acting is superb and the same goes for the sets and camera work. Come to think of it, the only thing I dislike (and greatly so) is the trailer as it to me profoundly fails to communicate the atmosphere of the movie.

The difference is that as a scientific reviewer you are not hiding a physical location and what you need is plausible deniability, which would still exist. In addition to this, actively attempting to deanonymise your reviewers is on the level of scientific misconduct that your employer and professional organisation should consider taking disciplinary action against you. I am not arguing that this makes it entirely safe to publish anonymised reviews and that we will not affect reviewer behaviour (maybe for the better in some cases, as "one-sentence reviews" will be something in the public record), but it is in stark contrast to the example that you bring up.

We are struggling badly with review quality in natural language processing though. Most likely due to the unprecedented expansion of the field over the last ten or so years. Reviewers are suffering with review loads far exceeding what one reasonably can manage mentally (used to be two to three papers per reviewer and now five would be considered rather generous). Authors and area chairs suffer from worse quality reviews due to reviewer inexperience and overload, not to mention how good reviewer/author correspondence with author and area chair comments frequently being ignored by the reviewers. To me, the last holdout of good peer review in the field is Transactions of the Association for Computational Linguistics (TACL), but there the acceptance bar is sky high compared to ACL Rolling Review (ARR) for better or worse.

The ACL leadership and senior members of the field are very much aware of this and are trying their best (ARR being an attempt to improve the situation, but I am unsure how much better it really is compared to the old system of conference reviewing now that we are a few years in). But there appear to be no easy fixes for a complicated, distributed system such as peer review. Every discussion I have with said leadership and other senior members always ends with us agreeing on the problems and likewise agreeing that despite considerable mental effort we are failing to come up with solutions.

Returning to the main topic. Nature is worthy of praise for making their peer review transparent and I say this as a massive Nature critic. It is a move I loved seeing from NeurIPS (then NIPS) and ICLR over a decade ago, as it helps younger researchers see what good (and bad) communication looks like and that even papers they know now are greatly appreciated received a fair amount of criticism (sometimes unwarranted). I have argued for ACL to introduce the same thing for nearly a decade at this point, but we still do not and I have never heard a solid argument as to why not (best argument was the technological effort, but OpenReview, with all its flaws, makes this even easier than with Softconf; not that it would have been that hard with Softconf either).

Marcus comes with a very standard cognitive science criticism of statistical approaches to artificial intelligence, many parts of which dates back to the late 50s from when the field was born and moved to distance itself from behaviourism. The worst part to me is not that his criticism is entirely wrong, but rather that it is obvious and yet peddled as something that those of us that develop statistical approaches are completely ignorant of. To make matters worse, instead of developing alternative approaches (like plenty of my colleagues in cognitive science do!), he simply reiterates pretty much the same points over and over and has done so at least for the last twenty or so years. He and others paint themselves as sceptics and bulwarks against the current hype (which I can assure you, I hate at least as much as they do). But, to me, they are cynics, not sceptics.

I try to maintain a positive and open mind of other researchers, but Marcus lost me pretty much at "first contact" when a student in the group who leaned towards cognitive science had us read "Deep Learning: A Critical Appraisal" by Marcus (2018) [1] back around when it was published. Finally I could get into the mind of this guy so many people were talking about! 27 pages and yet I learned next to nothing new as the criticism was just the same one we have heard for decades: "Statistical learning has limits! It may not lead to 'truly" intelligent machines!". Not only that, the whole piece consistently conflates deep learning and statistical learning for no reason at all, reads as if it was rushed (and not proofed), emphasises the author's research strongly rather than giving a broad overview, etc. In short, it is bad, very bad as a scientific piece. At times, I read short excerpts of an article Marcus has written and yet sadly it is pretty much the same thing all over again.

[1]: https://arxiv.org/abs/1801.00631

There is a horrible market to "sell" hype when it comes to artificial intelligence, but there is also a horrible market to "sell" anti-hype. Sadly, both brings traffic, attention, talk invitations, etc. Two largely unscientific tribes, that I personally would rather do without, with their own profiting gurus.

It was not so much the call for a complete halt that caused controversy, but rather this part of his piece in Time (my emphasis):

"Make it explicit in international diplomacy that preventing AI extinction scenarios is considered a priority above preventing a full nuclear exchange, and that allied nuclear countries are willing to run some risk of nuclear exchange if that's what it takes to reduce the risk of large AI training runs.

That's the kind of policy change that would cause my partner and I to hold each other, and say to each other that a miracle happened, and now there's a chance that maybe [our daughter] will live."