|
their cdn boxes are fbsd, most of the rest of the boxes (all the stuff on aws) are linux iirc
|
# ? May 3, 2020 12:57 |
|
|
# ? May 5, 2024 10:09 |
|
netcraft confirms bsd is dead
|
# ? May 3, 2020 13:21 |
|
freenas and pfsense are based on freebsd, as is playstation 3 and 4!
|
# ? May 3, 2020 20:54 |
|
Raere posted:freenas and pfsense are based on freebsd, as is playstation 3 and 4! freebsd is good but idk why you'd pick it over openbsd for a firewall
|
# ? May 3, 2020 21:36 |
|
what does bsd offer over linux?
|
# ? May 3, 2020 21:46 |
|
Ploft-shell crab posted:what does bsd offer over linux? openbsd is usually the first os to put third-party devs' nuts on the dresser for writing code with unsafe memory management also pledge and unveil are legit cool snd i hope other oses steal them
|
# ? May 3, 2020 21:55 |
|
you get to lord it over the insufferable greasy dude who brags about running ubuntu on his dell ultrabook "oh, that's cool dude. I run bsd on my vintage g3 powermac "
|
# ? May 3, 2020 22:05 |
|
plus i gotta say even though i've never run freebsd for a significant amount of time i really like their handbook
|
# ? May 3, 2020 22:14 |
|
the bsds (well, mostly openbsd) are often described as more secure* than your typical linux because of the extra security mechanisms and not introducing flashy new features * nothing is really secure
|
# ? May 3, 2020 23:02 |
|
can't get compromised by a usb stick if your os doesn't support usb
|
# ? May 4, 2020 01:20 |
|
thinking about getting one of these and loading a stable openbsd release on it now that syspatch is a thing. i ran a userland build on one of the older Edgerouters once and hoo boy
olives black fucked around with this message at 02:53 on May 4, 2020 |
# ? May 4, 2020 02:51 |
|
many years ago i was at a conference where someone asked theo why obsd didn’t use a memory safe language instead of their extreme efforts attempting to defend against C based attacks. Theo theod here was no suitable open source language, and at the time he was probably right Now: no one has written grep in Rust. QED. https://marc.info/?l=openbsd-misc&m=151233345723889 also, obsd grep recompiles really fast which supports my conviction that most cpu time used by the median bsd user is build world and is their primary optimization target
|
# ? May 4, 2020 13:39 |
|
PCjr sidecar posted:also, obsd grep recompiles really fast which supports my conviction that most cpu time used by the median bsd user is build world and is their primary optimization target de raadt says the quiet part loud elsewhere in that htread : https://marc.info/?l=openbsd-misc&m=151233345723889&w=2
|
# ? May 4, 2020 13:56 |
|
Ploft-shell crab posted:what does bsd offer over linux? absolutely nothing
|
# ? May 4, 2020 14:03 |
|
he’s right. compile times matter for development cycles. I’ve seen project build times spike up massively because a few rust components (low single digit percentage of the entire codebase) were introduced. also lol at rust cultists trying to pretend it’s a systems language.
|
# ? May 4, 2020 14:10 |
|
and when all your users are os devs or people cosplaying os devs that’s an important design consideration
|
# ? May 4, 2020 14:14 |
|
The Management posted:he’s right. compile times matter for development cycles. I’ve seen project build times spike up massively because a few rust components (low single digit percentage of the entire codebase) were introduced. could you please explain why rust ain't ready for prime time as a systems language? I have a friend who really wants me to put the k&r down and become a rustacean olives black fucked around with this message at 15:17 on May 4, 2020 |
# ? May 4, 2020 15:12 |
|
olives black posted:could you please explain why rust ain't ready for prime time as a systems language? I have a friend who really wants me to put the k&r down and become a rustacean My beef with rust is that the language spec isn't very stable. The core team changes it often enough that a bunch of packages require you to be on the 'unstable' version of the compiler, which implements an 'unstable' version of the language. Constant language tinkering also means that nobody wants to write docs or tutorials, because they know it'll be obsolete in a year all of this is the opposite of what you want for a systems language, which should be stable enough that you won't need to scrub your whole codebase every time somebody decides to change Rust's error-handling again like, I get it that the Rust devteam is a bunch of language nerds, and dicking around with language features is what sparks joy for them. it's also what holds back the language imo the situation is like C++ but without all the momentum that C++ picked up back when it wasn't changing every 3 years Poopernickel fucked around with this message at 15:30 on May 4, 2020 |
# ? May 4, 2020 15:25 |
|
Poopernickel posted:My beef with rust is that the language spec isn't very stable. The core team changes it often enough that a bunch of packages require you to be on the 'unstable' version of the compiler, which implements an 'unstable' version of the language. so p much all the stuff I've been telling this friend for years lol thank you
|
# ? May 4, 2020 15:59 |
|
olives black posted:could you please explain why rust ain't ready for prime time as a systems language? I have a friend who really wants me to put the k&r down and become a rustacean what that dude said. also the language is unsuitable for low level memory manipulation, which is often required for systems programs. if you can’t cast a struct to a different struct and then pick values out of it, or traverse a linked list, or deal with alignment issues, or access device registers in a predictable way, then your language isn’t useful for systems work. lol if his answer is going to be that you can do it with unsafe. also the religious fervor of rust enthusiasts is extremely annoying. it’s to the point where seeing rust on someone’s resumé is a red flag
|
# ? May 4, 2020 16:22 |
|
PCjr sidecar posted:many years ago i was at a conference where someone asked theo why obsd didn’t use a memory safe language instead of their extreme efforts attempting to defend against C based attacks. Theo theod here was no suitable open source language, and at the time he was probably right rust people should make their own os instead of trying to take another one over
|
# ? May 4, 2020 16:37 |
|
hifi posted:rust people should make their own os instead of trying to take another one over they would need a supercomputing cluster to compile it
|
# ? May 4, 2020 16:48 |
|
The Management posted:what that dude said. also the language is unsuitable for low level memory manipulation, which is often required for systems programs. if you can’t cast a struct to a different struct and then pick values out of it, or traverse a linked list, or deal with alignment issues, or access device registers in a predictable way, then your language isn’t useful for systems work. lol if his answer is going to be that you can do it with unsafe. is there really no way to just make a const pointer to volatile type at a given address in rust? i've seen 'embedded rust' thrown around but never actually looked into it, how the gently caress do they access sfrs otherwise?
|
# ? May 4, 2020 17:25 |
|
PCjr sidecar posted:Now: no one has written grep in Rust. QED. whats ripgrep precious
|
# ? May 4, 2020 19:54 |
|
Progressive JPEG posted:whats ripgrep precious "not grep"
|
# ? May 4, 2020 20:02 |
|
it's not a unix tool because it's pleasant to use
|
# ? May 4, 2020 20:06 |
|
Progressive JPEG posted:it's not a unix tool because it's pleasant to use but yeah rg is a lifesaver and i use it like every day to deal with my many repos
|
# ? May 4, 2020 20:30 |
|
this is the future obsd users want https://twitter.com/ao_kenji/status/1257267898306801665?s=21
|
# ? May 5, 2020 02:36 |
|
PCjr sidecar posted:this is the future obsd users want https://twitter.com/ao_kenji/status/1257267898306801665?s=21 theo says the older machines help them find bugs
|
# ? May 5, 2020 02:52 |
|
til m88k was a thing
|
# ? May 5, 2020 04:00 |
|
olives black posted:what would you have used instead lisp obviously
|
# ? May 5, 2020 04:42 |
|
olives black posted:what's the best guide for reading about OpenBSD kernel stuff? I know that FreeBSD has its own well-regarded book but that obvs doesn't apply here the source code is the best guide OP
|
# ? May 5, 2020 04:43 |
|
olives black posted:openbsd 6.7 just hit beta aw drat I shoulda done the rest of the work to get mvme88k ported all the way forward so it could be brought back with 6.7 maybe for 7.0
|
# ? May 5, 2020 04:46 |
|
PCjr sidecar posted:this is the future obsd users want https://twitter.com/ao_kenji/status/1257267898306801665?s=21 they wouldn’t be the last working ones if the couple people I know with buried LUNA-88K systems would unbury them and send them my way then I could help bring back NetBSD for m88k OpenBSD makes all sorts of claims about security but one of the worst things for it is code duplication, of which they have a lot, because they refuse to use a modular kernel—to the point where most of the drivers are duplicated between similar systems (like mvme68k/mvme88k/aviion/luna88k) building a newer OpenBSD for mvme88k meant resurrecting deleted files, applying the same changes to them that were made to nearly identical files elsewhere in the codebase, building, and rebooting with the new kernel it was cool that it just worked but, like, why were those drivers even different, that’s stupid and wrong
|
# ? May 5, 2020 04:57 |
|
Raere posted:til m88k was a thing it was a good RISC CPU and at one point the part of the workstation & server industry that didn’t make its own CPUs promoted it as “the future” to the point of having a binary compatibility standard via an industry consortium with the unfortunate name “88Open” but there were only two models ever produced, 88100 and 88110, and after that Motorola was all PowerPC and most of the 88K users switched to other platforms (like Data General switching to Intel in later AViiON workstations and servers and NCD & Tektronix switching to MIPS in their X terminals) the PowerPC 601 was essentially a drop-in replacement—at least electrically/design-wise, if not pin-compatible—for the 88110 so that the companies who had been working on 88K systems could transition to PowerPC easily Motorola still produced the 88K into the 2000s though, including VME CPU & memory boards, because it wound up used in a couple important niches like military avionics
|
# ? May 5, 2020 05:05 |
|
olives black posted:openbsd 6.7 just hit beta for gently caress's sake 2
|
# ? May 5, 2020 14:34 |
|
eschaton posted:they wouldn’t be the last working ones if the couple people I know with buried LUNA-88K systems would unbury them and send them my way my guess is that it's because there's still differences between those systems and they want to be able to easily make and test changes to one without messing up the others?
|
# ? May 5, 2020 14:51 |
|
carry on then posted:for gently caress's sake 2
|
# ? May 5, 2020 15:53 |
|
what is the smallest prime number?carry on then posted:for gently caress's sake 2
|
# ? May 5, 2020 17:08 |
|
|
# ? May 5, 2024 10:09 |
|
p good review of OpenBSD's mitigations https://youtu.be/3E9ga-CylWQ his goal was obviously to take the piss but he wound up having mostly nice things to say throughout
|
# ? May 5, 2020 17:38 |