|
I haven’t worked on testing since about then but I’m still plenty associated with it in folks minds your X stuff is so memorable in part because of just how clear it is for something that’s often explained pretty opaquely, and then there are the demos
|
#
?
Jun 8, 2019 02:52
|
|
|
|
| # ? May 26, 2024 20:02 |
|
|
Suspicious Dish posted:Because it was genuinely more fun than what I was doing at work that day. Speakings of, my new job lasted all of six months. where're you off to next?
|
|
#
?
Jun 8, 2019 04:12
|
|
|
Beamed posted:where're you off to next? heading back into games
|
|
#
?
Jun 8, 2019 05:45
|
|
|
hey, now that Telltale is dead and can't possibly care about secrecy, what were the Linux ports you were working on even for?
|
|
#
?
Jun 8, 2019 07:25
|
|
|
I wish Monty would do some more explainer videos, those were real good.
|
|
#
?
Jun 8, 2019 08:06
|
|
|
pseudorandom name posted:hey, now that Telltale is dead and can't possibly care about secrecy, what were the Linux ports you were working on even for? if they were secret and weren’t for steam it’s a 99.999% chance it was stadia
|
|
#
?
Jun 8, 2019 13:08
|
|
|
pseudorandom name posted:hey, now that Telltale is dead and can't possibly care about secrecy, what were the Linux ports you were working on even for? https://www.dslreports.com/forum/r31704565-X1-Walking-Dead-game-on-X1 The_Franz posted:if they were secret and weren’t for steam it’s a 99.999% chance it was stadia we had a stadia project in the works but company died before it got out of prepro
|
|
#
?
Jun 8, 2019 15:28
|
|
|
Rufus Ping posted:He said "a" though. He wanted the singular and got it right You can't stack one sheet of lasagna. By nbsd's rules he wanted 'lasagnas'
|
|
#
?
Jun 8, 2019 17:15
|
|
|
freeing stuck post from radiums clutches e: hm there wasn't one, oh well
|
|
#
?
Jun 8, 2019 17:47
|
|
|
I just installed NixOS; partly because I wanted to learn a new computer thing, and partly because I was nostalgic for the evenings I spent as a teenager trying to tweak my Gentoo system. It's not at all like it used to be. Both sound and GPU drivers worked out of the box. Only bit of excitement was that KDE (the suggested default) would crash upon login, but that was easily fixed by switching to GNOME.
|
|
#
?
Jun 8, 2019 21:07
|
|
|
i'm starting to think the real reason linux people love scrapping everything and starting over is now they've had to endure the horrors of actually using a computer
|
|
#
?
Jun 8, 2019 21:21
|
|
|
Athas posted:I just installed NixOS; partly because I wanted to learn a new computer thing, and partly because I was nostalgic for the evenings I spent as a teenager trying to tweak my Gentoo system. nixos is cj as gently caress but its a v diff cj than gentoo. if anything its the anti-gentoo. it's for cjs who care about having solved problems rather than relishing the act of solving identical problems over and over in an eternal simulacrum of 2003 i wouldnt want to run nixos on my workstation but i have it for small server i was always rebuilding and i literally havent thought about it or touched it in over a year. it does attract gentoo/slashdot types because of some weird torch-passing reputation but they tend to bounce quickly once they realize nixos has wholly embraced systemd and has no interest in changing
|
|
#
?
Jun 9, 2019 01:52
|
|
|
i feel that way about arch. it doesn't come with everything pre-installed but all the packages work fine out of the box with reasonable defaults plus pretty consistently good documentation via the wiki, and the packages are kept up to date with no or minimal patching from the distro it does effectively require already knowing how to do a linux and how the pieces fit together to get what you want, but if you're already starting with that knowledge level then it's pretty much set and forget ive had a ThinkPad running the same arch environment for something like three years now. if it was on ubuntu or something i'd have probably needed to reformat once every 6-8 months in that timespan either due to new distro releases or from something simply making GBS threads itself in a way that isn't worth diagnosing. instead i just "pacman -Syu" every couple months or so and bump everything to latest and it all just kinda works as installed
|
|
#
?
Jun 9, 2019 02:53
|
|
|
arch is like crossfit... you do a lot of work, and you get real sweaty and tired. but in the end, you're not really accomplishing anything.
|
|
#
?
Jun 9, 2019 02:55
|
|
|
For the last couple of nights I've been setting up Arch on an old box I had lying around. My job requires me to lightly admin a bunch of linux servers but I never really have to do much, so I've embarked on this Arch experiment as a way of actually learning some stuff about linux. For that its been quite nice, and the wiki is a surprisingly well-written resource. I probably wouldn't run it if I had to actually use it to do a job or anything productive though.
|
|
#
?
Jun 9, 2019 03:13
|
|
|
pacman is the best package manager, it's kind of shocking how bad everything else is by comparison
|
|
#
?
Jun 9, 2019 04:13
|
|
|
Red Hat continually hosed themselves over with that one. rpm still uses bdb (yes, that bdb, the text file that is less stable that csv and shouldn't be called a "database") under the hood because... erm, nobody knows why. So more layers on top it goes. razor was a good idea but team politics means it had to be cut. apt is just... I really don't know how you invent something as broken as apt.
|
|
#
?
Jun 9, 2019 05:00
|
|
|
Suspicious Dish posted:apt is just... I really don't know how you invent something as broken as apt. second-system effect after the dselect fiasco
|
|
#
?
Jun 9, 2019 05:15
|
|
|
when i have to use npm at work, I long for something as stable and fast as apt
|
|
#
?
Jun 9, 2019 05:47
|
|
|
akadajet posted:arch is like crossfit... thats the weird part tho, ive found arch to need little to no janitoring effort in practice, which is absolutely out of character with how niche distros usually work its like someone made sure that any janitoring you do should be as predictable and straightforward as possible, so that on the occasion you want to change something it's done after 5-10 minutes of following a surprisingly competently written wiki page, and otherwise it's basically in a steady state of periodically updating packages and not having anything spontaneously break on you i still wouldnt use it for something im getting paid to do (lol at being the arch guy at work) but its very pleasant for personal machines
|
|
#
?
Jun 9, 2019 05:57
|
|
|
arch requires a decent amount of janitoring to set up, especially if you need something less than standard. encrypted disk for work laptops has been an enjoyable hike through several contradictory wiki articles the last few times ive done it. the instructions even for standard installs are still pretty iffy. after getting a system to an installed and usable desktop state i'd say it's more painless than most systems: dist-upgrade and equivalents have typically caused me more pain than rolling release, even with the occasional breaking change/manual action needed. pacman and aur integration tools are still amazing regardless.
|
|
#
?
Jun 9, 2019 08:07
|
|
|
actually now that you mention it, I do specifically remember getting disk encryption set up the first time being a bit of trial and error, but I hadn't needed to touch it since that initial setup years ago and had forgotten about it
|
|
#
?
Jun 9, 2019 08:51
|
|
|
In my experience Arch needs constant attention. When doing a simple Pacman -Syu you really need to pay attention if anything has changed and make the required changes. Often these are well documented in the wiki, but if you don't follow the instructions you're hosed. I have this old laptop that I put Arch on just to okay around with it and I would use it for a while then put it away for a couple of weeks. That is really dangerous. Getting an arch system updated after a couple weeks of not using it is very painful. In those weeks they might have changed several important things and if you just go in and run Pacman you'll gently caress up your system. Instead, you have to spend an hour reading the wiki and figuring out what changed and make a plan on which updates to do, what manual changes they need, and most importantly in what order. This has been my experience with Arch. If you give it constant attention it's fine. Treat it like any other Linux and you'll be punished for it. And I just cannot be bothered putting in that kind of effort. I just wanna apt update && apt upgrade -y or dnf upgrade -y
|
|
#
?
Jun 9, 2019 09:27
|
|
|
ohh western digital is adding zoned device support (smr) to btrfs, cool is it actually possible to get a drive with smr as a consumer, or are those drives still sold by the pallet to cloud providers?
|
|
#
?
Jun 9, 2019 10:34
|
|
|
spankmeister posted:In my experience Arch needs constant attention. When doing a simple Pacman -Syu you really need to pay attention if anything has changed and make the required changes. Often these are well documented in the wiki, but if you don't follow the instructions you're hosed. the most I've experienced over the last few years of using it is if the kernel gets updated and I don't reboot afterwards then mounting new volumes sometimes won't work until I do that. it's otherwise been set and forget
|
|
#
?
Jun 9, 2019 11:23
|
|
|
the thing that seems to be missing from all the linux package managers is a robust mechanism for doing stepped upgrades if you are way behind on updates you should have the option to sequentially update to intermediate versions in the order they were released instead of going directly to the latest version, so that you can follow the most well tested path i guess it would mean that repositories have to keep outdated versions of packages available for longer but i think it would be worth it to avoid breaking systems that are slow to update
|
|
#
?
Jun 9, 2019 11:44
|
|
|
in gentoo you can have the system only install packages marked as stable, but also mark specific packages as unstable, maybe thats up your alley suse might be a candidate too, it has that zypper package management snapshot-thing integrated (using btrfs) which makes it really easy to roll back stuff
|
|
#
?
Jun 9, 2019 11:56
|
|
|
I had only been running NixOS for a few hours before I created my first pull request, and today I made another. While it's clearly a cj-heavy distribution (because it has so few users and I have obscure needs), it does feel fundamentally sane and approachable. In contrast, I ran Debian for a decade and never managed to put together a package for upstream.
|
|
#
?
Jun 9, 2019 14:09
|
|
|
Athas posted:I had only been running NixOS for a few hours before I created my first pull request, and today I made another. While it's clearly a cj-heavy distribution (because it has so few users and I have obscure needs), it does feel fundamentally sane and approachable. In contrast, I ran Debian for a decade and never managed to put together a package for upstream. the only way to create a package for Debian is to create an open source project and wait. eventually a random build at least 6 months old will appear in Debian unstable with some strange changes applied to it
|
|
#
?
Jun 9, 2019 16:34
|
|
|
gnu hyperbole is the poo poo. i probably made fun of it 10 years ago but i'm getting all sorts of ideas with it.
|
|
#
?
Jun 9, 2019 22:50
|
|
|
eschaton posted:lasagñæ lmao
|
|
#
?
Jun 10, 2019 00:31
|
|
|
hobbesmaster posted:the only way to create a package for Debian is to create an open source project and wait. eventually a random build at least 6 months old will appear in Debian unstable with some strange changes applied to it why does this encryption code need random numbers anyway
|
|
#
?
Jun 10, 2019 01:50
|
|
|
Progressive JPEG posted:why does this encryption code need random numbers anyway i'mma be real here, since i've talked a whole bunch of poo poo about debian on this very forum because of that issue actually reading about this revealed that upstream openssh key generation program was using stack garbage as a significant source of entropy, in the sense that patching out the use of heap garbage reduced the entropy bits to a brute-forceable level i'm not sure the entire fault lies with debian's maintainers in this particular situation edit: to clarify, ssh key generation falls under the classification of "long-lived keys". the gold standard for this sort of key generation is to consider /dev/urandom to be of insufficient quality and requiring that /dev/random be used. stack garbage is uh, of somewhat lower-quality randomness than /dev/urandom. Sapozhnik fucked around with this message at 02:00 on Jun 10, 2019 |
|
#
?
Jun 10, 2019 01:58
|
|
|
Sapozhnik posted:the gold standard for this sort of key generation is to consider /dev/urandom to be of insufficient quality and requiring that /dev/random be used. and for what it's worth, I think the best crypto people keep telling everyone to stop saying this, urandom is fine for everything, and random is dumb paranoid garbage and not in the good way.
|
|
#
?
Jun 10, 2019 02:40
|
|
|
crazypenguin posted:and for what it's worth, I think the best crypto people keep telling everyone to stop saying this, urandom is fine for everything, and random is dumb paranoid garbage and not in the good way. idk if /dev/random is good enough to seed daughter prngs or not, but i do know that relying on /dev/random is a good way to hardlock your application on dozens of nodes at the same time
|
|
#
?
Jun 10, 2019 03:10
|
|
|
Yeah once you have enough bits of entropy needed to key an hmac csprng then for any sort of remotely realistic crypto applications you are good to go. But that's just my non professional opinion.
|
|
#
?
Jun 10, 2019 04:54
|
|
|
Sapozhnik posted:i'mma be real here, since i've talked a whole bunch of poo poo about debian on this very forum because of that issue No, that's a mischaracterization of the issue. There were three entropy sources: /dev/random Process PID Uninitialized stack garbage The Debian maintainer attempted to remove 3, but because of a misunderstanding of the code, accidentally removed both 1 and 3, leaving only the PID as the only RNG source.
|
|
#
?
Jun 10, 2019 07:48
|
|
|
Just use 7 IMO
|
|
#
?
Jun 10, 2019 08:08
|
|
|
Suspicious Dish posted:No, that's a mischaracterization of the issue. There were three entropy sources: The reason was very dumb. As I recall the maintainer did this because they just wanted to get rid of compiler warnings about use of uninitialized values without understanding why those warnings were generated.
|
|
#
?
Jun 10, 2019 08:19
|
|
|
|
| # ? May 26, 2024 20:02 |
|
|
Suspicious Dish posted:No, that's a mischaracterization of the issue. There were three entropy sources: what's super bad about this is that crypto 101 courses teach that when you mix (xor) two entropy sources, one bad and one good, you get back entropy at least as good as the good source. bad entropy cannot harm good, it can only improve it. although i seem to remember the debian maintainer did it just because they saw a compiler warning about use of uninitialized data? and thus didn't even think of it in terms of removing a bad entropy source, just thought they were fixing a bug. still points to "they should not have been allowed to touch that poo poo," because random noobs touching crypto code is just... idk how to describe it, you just don't fuckin want people who aren't wizards touching a single byte of crypto related source without expert supervision (fake edit: beaten by spankmeister)
|
|
#
?
Jun 10, 2019 08:30
|
|
