Year of the OpenBSD desktop | Hacker News Reader

Year of the OpenBSD desktop

blog.tintagel.pl

60 pointstdurden21 comments

I think the biggest takeaway from this is the amount of poorly thought out software that exists there which assumes that Linux is the only unix. From the link on page about porting Dart -- https://github.com/dart-lang/sdk/issues/10260 -- people have gotten it working, but upstream is completely silent on porting patches and working with people on getting the required alterations in-tree. Even starting with the basics, like the shell interpreter, too much software assumes that bash is part of the base user space, instead of a package that will be somewhere other than /bin. It just feels that after Linux hit a certain amount of market share, too many developers decided that portable well-written code wasn't worth it, and ignored the millions of systems that run other operating system, to the detriment of everyone.

"I think the biggest takeaway from this is the amount of poorly thought out software that exists there which assumes that Linux is the only unix."

That's true. The biggest takeaway, though, is the assumption that people will care to port mainstream software to OpenBSD when virtually nobody using that software runs OpenBSD & their community discourages amateur programmers. It's not a wise expectation. Instead, they should expect to have to port it all themselves to their particular OS of choice while being thankful when people do it for them with great portability.

In that case, you'd have a bunch of semi-functional desktops out there running OpenBSD with a ton of features and options on Linux desktops. (Pauses.) Exactly what happened.

> That's true. The biggest takeaway, though, is the assumption that people will care to port mainstream software to OpenBSD when virtually nobody using that software runs OpenBSD & their community discourages amateur programmers. It's not a wise expectation. Instead, they should expect to have to port it all themselves to their particular OS of choice while being thankful when people do it for them with great portability.

You got it a bit wrong or didn't read the linked github issue in detail. I run OpenBSD and I ported Dart to OpenBSD. I tried to work with upstream (Dart team at Google) to incorporate my changes or work on a way that allows them to take the changes and for me to maintain the code long term. In this specific case the upstream works behind closed doors which makes it hard to cooperate on things like this.

Oh, I read it right:

" I ported Dart to OpenBSD. I tried to work with upstream (Dart team at Google) to incorporate my changes or work on a way that allows them to take the changes and for me to maintain the code long term. "

You wanted them to work so that your unpopular OS is supported. My claim was that you'd have to fork it and do the work yourself in many cases since the OS project is for benefit of you and small niche of developers/users. The irony is that comments like that deride non-OpenBSD FOSS for not going through trouble of OpenBSD support while the LibreSSL commits mocked and stripped OpenSSL's support of VMS, etc. as crap that's not worth the effort users be damned. Same attitude in place.

The Dart team is probably targeting Windows, Mac, and Linux (w/ Android). Those are THE desktop OS's in terms of uptake and capabilities available. Adding PC-BSD, OpenBSD, eComstation OS/2, Syllable, BeOS, MorphOS, MenuetOS... these could benefit someone but represents lots of work with almost no gain due to near-zero market share. So, for niche projects, it's on you to get the work done if your goals and theirs aren't compatible. They're doing nothing wrong even if a bit non-optimal given benefits of portable code tested on many toolchains.

> I think the biggest takeaway from this is the amount of poorly thought out software that exists there which assumes that Linux is the only unix.

If shit continues like it has been as of late, it may well be that they start thinking Fedora is the end all, be all, of unix...

Portablity isn't even hard. Some maintainers just have a tendency to get huffy and say "but I've always used Linux, so all Linuxisms are correct" - I think this tendency derives from GNU's entirely deliberate use of embrace-and-extend tactics with its ecosystem of GPLed code.

But that's OK. We have a ports tree, we don't need upstream.

OpenBSD has been my full-time desktop at work since 2011, and I used it intermittently before that.

I have also used it on laptops but you need to be a bit selective on what you try for those. OpenBSD does not include any "proprietary blobs" so devices that don't have open source drivers generally aren't supported.

I love its rock-solid stability, consistent and sane configuration, easy installation and low frustration.

It's not what you want if you're seeking to wring every bit of performance from your hardware, but it keeps improving and it's always been more than adequate for me.

Sad Netflix is missing. The web should be standard and identical everywhere, which means Firefox and Chrome et al should operate the same everywhere.

There was just a comment by someone in a recent thread about apps in browsers being the size of the original Doom, which stated that the web is the only real, true cross-platform production development and consumption environment (looking for it) and it's almost true... Except for where lobbying is strong enough to even break the open, standard web with proprietary software.

Boo.

Edit: found the comment, by @dasil003 https://news.ycombinator.com/item?id=11552598

Edit 2: despite the contentious conversation dasil003's comment caused, I agree with it, but you could already tell that.

Should be, perhaps. Much like Java should be standard and identical everywhere. Differences are inevitable, priorities depend a lot more on real world constraints than fleeting desires. Those proprietary standards exist because the web continues to be rather terrible if your problem's not easily explicated in javascript.

The Javascript monoculture that exemplifies the modern web is perhaps the greatest thing holding back good cross-platform development. There's no way to experiment with other languages on equal footing, there's no way to easily extend browsers to provide hooks to native libraries and functionality in a reasonable manner. If the web is the only cross platform environment, then give me my walled gardens, because they're gonna be the only place open enough to actually get things done. The web's a greenhouse, with only the illusion of freedom until you get next to the glass walls.

> The Javascript monoculture that exemplifies the modern web is perhaps the greatest thing holding back good cross-platform development.

I dunno, there was a time when javascript wasn't the only game in town, and it sucked. ActiveX controls for hooking into native libraries, Java applets that assumed you had a JVM installed, subtle differences between javascript and jscript, VBS in the browser.. It was terrible, and the web is better now that they are largely a thing of the past.

I'm not a superfan of javascript, but building cross platform stuff in the browser seems easier now that you can count on its ubiquity and everyone is generally on board. Getting things done might be easier in walled gardens, but those gardens aren't usually cross platform.

Basically, it's a real desktop once it can be used for a limited set of day-to-day activities. By that definition, it's the year of FreeDOS and Atari desktops. Yay!

Edit to add: The Year of the Amstrad 8-bit desktop!

http://www.symbos.de/

I actually didn't like the post, I hate the notion that one person using something means others should, but that's besides the point.

I use openbsd daily as my work machine, and sure, it has some quirks (chrome used to lag something fierce and it's still slow even now, wifi will randomly start lagging out and dropping packets etc;)

While I personally can get around all day every day on openbsd with only marginal problems, I wouldn't push it on everyone like I did with linux, because while I can do everything I want to do with openbsd I'm fully aware it can't do the more 'advanced' things which have just made it over to linux (games, for example).

and that's ok. I'm definitely not crippled or even hindered by using it, so it's not comparable to freedos in my opinion.

Good assessment. To be clear, the comparison is kind of narrow. The claim is that one person (or a few) functioning in their use case day to day on it means it's "ready for the desktop." Remember what desktop we ran businesses and home stuff on before Windows? And which version has been updated probably the most? ;)

Many Fortune 500 companies run on terminal apps to a large degree even if many put GUI's on front of them. So do a lot of warehouses and retail stores. Doesn't really hinder them day-to-day. Just not convenient, flexible, and so on as modern IT. FreeDOS is certainly not comparable to OpenBSD or anything but seems to show the folly of "someone uses it day to day so let's call it a real desktop and ready!" Suddenly, the word desktop gets much less meaningful really quickly.

> I actually didn't like the post, I hate the notion that one person using something means others should, but that's besides the point.

I'm sorry if it came out that way. The intention of the post was totally opposite. I was referring to the fact that Linux on the desktop already happened for a lot of people (including me) and that OpenBSD for a lot of people is already on the desktop in the same way. I'm happy with people using whatever system suits them, the whole thing was just a comment on recent tweets & comments online about 2016 being the year of the OpenBSD desktop.

Already said why it was inaccurate. That said, had it just been "Whose using OpenBSD on the Desktop!?," then it would've made a good write-up and list. I still enjoyed seeing the list and your community's little triumphs of having the desktop despite the challenges you all face getting there. I respect that. :)

Do releases receive security updates on packages? Looking at package list it seems all packages have dates going back to version release. I'm not sure superior mitigation technology and auditing necessarily outweigh 6 months hackers have on outdated version of a package.

They are provided by M:Tier. See https://stable.mtier.org/ and http://www.mtier.org/index.php/solutions/apps/openup/

Really wish there was official channel for this. Has OpenBSD team said anything about M:Tier?

Ports receive security updates on the -stable branch in CVS, but the project does not provide updated binary packages for the -stable branch after the release. M:Tier provides updated binary packages, or you can always just update the -stable ports branch and build/install the updated ports yourself.

That's a half-truth. Ideally they would, but they don't.

vlc 2.2.1 (CVE-2015-5949)

php 5.6.18 (CVE-2016-3142, CVE-2016-3141)

firefox 44.0.2 (CVE-2016-1969, CVE-2016-1977, CVE-2016-2790, CVE-2016-2791, CVE-2016-2792, CVE-2016-2793, CVE-2016-2794, CVE-2016-2795, CVE-2016-2796, CVE-2016-2797, CVE-2016-2798, CVE-2016-2799, CVE-2016-2800, CVE-2016-2801, CVE-2016-2802, CVE-2016-1979, CVE-2016-1950, CVE-2016-1974, CVE-2016-1973, CVE-2016-1970, CVE-2016-1971, CVE-2016-1975, CVE-2016-1976, CVE-2016-1972, CVE-2016-1966, CVE-2016-1968, CVE-2016-1967, CVE-2016-1965, CVE-2016-1964, CVE-2016-1963, CVE-2016-1962, CVE-2016-1961, CVE-2016-1960, CVE-2016-1959, CVE-2016-1958, CVE-2016-1957, CVE-2016-1956, CVE-2016-1955, CVE-2016-1954, CVE-2016-1953, CVE-2016-1952)

Obviously these are cherry-picked and are historically known for their many vulns; but they are also popular enough that one would assume that they'd get special attention.

Depending on ports from -STABLE in a security sensitive environment is a bad idea imo.

It should be noted that "The packages and ports collection does NOT go through the same thorough security audit that is performed on the OpenBSD base system." -- http://www.openbsd.org/faq/faq15.html

Reply on news.ycombinator.com