|
OldAlias posted:do you use either of these as a desktop operating system? packages are outdated as gently caress. this matters a lot less often than you might think centos 7 is a perfectly adequate desktop
|
#
?
Jul 2, 2016 08:24
|
|
|
|
| # ? Apr 29, 2024 13:22 |
|
|
Soricidus posted:just use fedora, it's the least worst desktop distro and you're gaining experience wiht the things rhel and centos are going to have in a few years. using anything else is just silly. yeah this is exactly what I do Notorious b.s.d. posted:i would argue that fedora is more of an advertisement for redhat than a qa process that's fine, redhat is good and Good
|
|
#
?
Jul 2, 2016 10:11
|
|
|
packages being old would only matter if there were actual progress on the linux desktop, as it stands you will just have vintage flaws rather than some new vulgar ones
|
|
#
?
Jul 2, 2016 10:21
|
|
|
fedora wasn't fit to fly on a laptop i was setting up. that and the name makes me think of a dweeb tipping his hat.
|
|
#
?
Jul 2, 2016 12:53
|
|
|
m'linux
|
|
#
?
Jul 2, 2016 14:38
|
|
|
Silver Alicorn posted:m'linux
|
|
#
?
Jul 2, 2016 14:58
|
|
|
Notorious b.s.d. posted:sip/rtp over openvpn is a gigantic bag of hosed. you are completely, totally screwed. they don't support ipsec, they're all yealink desk and conference phones, what we've tested so far was a conference phone and a desk phone communicating over the vpn they just use .tar files with a config file and verious ssh keys and user/pass auth file but it does simplify deployment a bit over trying to use nat like we were, current number of phones is around 100-150 i think. feels like if i wanted to do something similar with a branch office i would want some sort of hardware vpn instead of having every phone connect to it themselves?
|
|
#
?
Jul 2, 2016 18:38
|
|
|
OpenVPN/UDP would be equally as affective as IPsec, it's only the startup costs that are higher.
|
|
#
?
Jul 2, 2016 19:29
|
|
|
MrMoo posted:OpenVPN/UDP would be equally as affective as IPsec, it's only the startup costs that are higher. probably not the virtue of ipsec for voip is that you can avoid tunneling. your packet stream doesn't have to get re-assembled or re-fragmented in another packet stream. your packets just get encrypted as they travel along. poof. openvpn over udp is much better than openvpn over tcp, but it is still a big bag of hurt for latency/jitter-sensitive apps
|
|
#
?
Jul 3, 2016 16:49
|
|
|
Apocadall posted:they don't support ipsec, they're all yealink desk and conference phones, what we've tested so far was a conference phone and a desk phone communicating over the vpn a vpn tunnel will be 100% fine for the provisioning. it will get kinda hairy for the actual voip traffic. if you can afford hardware devices you can sidestep this problem and just set up ipsec. the devices on either end (the phones and the phone server) won't need to know anything is happening. the branch office router will just transparently encrypt packets as they go. that said, you will probably need to use public ip space on your office networks to make it 100% transparent and avoid gre tunneling. (have i ever mentioned that i hate working with voip?)
|
|
#
?
Jul 3, 2016 16:53
|
|
|
p.s. the symptom of voip being hosed by your branch office vpn is that it works just fine 90% of the time then 10% of the time you either get hilarious delays or broken speech it's really frustrating to test or troubleshoot
|
|
#
?
Jul 3, 2016 16:55
|
|
|
it is weird how important network protocols and software seems to keep getting designed with people who do not understand the basic problems on the lower couple of layers, anyone who tunnels a tcp-style stream over another tcp-style stream ought to have some basic understanding of what they are getting themselves into i mean, i can't design that poo poo well, but it certainly seems like people keep falling into traps which were understood 40 years ago
|
|
#
?
Jul 3, 2016 17:33
|
|
|
i don't know how nat + http proxies became the default choices for office networks but boy howdy was that the dumbest thing ever to happen to this industry dumber than windows
|
|
#
?
Jul 3, 2016 17:39
|
|
|
Notorious b.s.d. posted:i don't know how nat + http proxies became the default choices for office networks but boy howdy was that the dumbest thing ever to happen to this industry what would you prefer then? (legit question)
|
|
#
?
Jul 3, 2016 21:23
|
|
|
Shinku ABOOKEN posted:what would you prefer then? (legit question) ipv6 public IP for life
|
|
#
?
Jul 3, 2016 23:41
|
|
|
Notorious b.s.d. posted:i don't know how nat + http proxies became the default choices for office networks but boy howdy was that the dumbest thing ever to happen to this industry even 20 years ago IPv4 space was well understood to be scarce, and regular ISPs didn't want to provision sites with subnets etc. when they could just provision a T1 or DSL or cable modem with a single IP address and leave everything else up to the customer plus NAT could do double duty as a firewall, and a proxy could improve performance via caching, or be used to monitor those unruly minions and block access to things the boss doesn't like quote:dumber than windows that's a bit of a stretch
|
|
#
?
Jul 4, 2016 00:04
|
|
|
the fun part of nat + proxy is they are usually not configured correctly or entire vlans are going around them. seeing this way to much lately
|
|
#
?
Jul 4, 2016 00:15
|
|
|
as a security nut, it boggles the mind as to why open source software sets default nat proxies ekiga and pidgin, I'm looking at you I mean, pidgin has default file trwnsfer proxies. Really?? That poo poo should just be pushed as defaults by the server and overridden by the user.
|
|
#
?
Jul 4, 2016 00:44
|
|
|
fortunately us radio hams can still be smug about owning 44.*
|
|
#
?
Jul 4, 2016 00:48
|
|
|
Apocadall posted:because it's actually made to be an enterprise workstation and not some bespoke custom bullshit that's useless outside of toy applications? i mean i know why fedora is better but i just needed a link to send to someone whos too boneheaded to use anything other than sabayon (is this a gaming linux or something)
|
|
#
?
Jul 4, 2016 03:15
|
|
|
Korean Boomhauer posted:i mean i know why fedora is better but i just needed a link to send to someone whos too boneheaded to use anything other than sabayon (is this a gaming linux or something) comedy option: steam OS
|
|
#
?
Jul 4, 2016 03:54
|
|
|
Really, he should use the OS that is most frequently tested for his games. Which is probably some flavor of ubuntu.
|
|
#
?
Jul 4, 2016 03:56
|
|
|
oh god, it's based on gentoo
|
|
#
?
Jul 4, 2016 03:58
|
|
|
the package system is called "entropy" that seems appropriate
|
|
#
?
Jul 4, 2016 07:01
|
|
|
eschaton posted:even 20 years ago IPv4 space was well understood to be scarce, and regular ISPs didn't want to provision sites with subnets etc. when they could just provision a T1 or DSL or cable modem with a single IP address and leave everything else up to the customer twenty years ago dsl and cable were unspeakably rare, and ip space was both common and essentially cost free i know guys who still have /24s allocated to them personally. arin gave zero fucks in the 90s edit: also cable didn't leave anything up to the customer in the early days. the first cable connection i ever used came with a cisco access router and a guy to configure it. i poo poo you not. this was, needless to say, very expensive
|
|
#
?
Jul 4, 2016 07:02
|
|
|
Notorious b.s.d. posted:i don't know how nat + http proxies became the default choices for office networks but boy howdy was that the dumbest thing ever to happen to this industry Blaster
|
|
#
?
Jul 4, 2016 11:03
|
|
|
https://www.bassi.io/articles/2016/07/05/gsk-demystified-1/ https://blogs.gnome.org/hughsie/2016/07/05/flatpak-and-gnome-software/ gnome continuing to own pretty hard today, also making big strides towards being a respectable desktop environment and rendering distributions less relevant over time
|
|
#
?
Jul 5, 2016 16:37
|
|
|
|
|
#
?
Jul 5, 2016 16:50
|
|
|
is flatpak the thing where you make distros irrelevant by installing them all at the same time in hidden directories, or did they come up with a less hilariously bad way of doing it
|
|
#
?
Jul 5, 2016 18:29
|
|
|
Soricidus posted:is flatpak the thing where you make distros irrelevant by installing them all at the same time in hidden directories No, although it does kinda seem that way at first glance. The idea is that GNOME, KDE, and Freedesktop distribute "runtimes", which are collections of libraries typically required by the apps that run on their respective desktop environments, and they release a new runtime every time they release a new version of their desktop environments. Also those runtimes are built and patched by the desktop projects themselves, not by RH or Debian or whatever. A runtime will contain things like Gtk+, Wayland, and PulseAudio client libraries, but not Xorg or GNOME Shell or whatever. Those are considered part of the OS and are not managed by FlatPak, so the runtimes are not distros. FlatPak apps run in containers. The runtime gets bind-mounted into /usr, the app gets mounted into /app (with a similar internal structure to /usr), and each app gets a private /var that ends up somewhere under the user's ~/.local folder. Apps can also get the user's home directory mounted into the container, but the preferred way for them to interact with your files is for them to request files using DBus services which implicitly grant access to a user's files by displaying a File Open dialog and then passing the app an fd for whatever file the user picked. The whole thing is also modularized nicely; it stores all the installed files in OSTree which is a Git type thing for binaries that can perform side-by-side runtime installations and upgrades efficiently in both space and bandwidth, and the actual containerization is done by a reusable thing called BubbleWrap.
|
|
#
?
Jul 5, 2016 19:22
|
|
|
thanks for the effortpost, that sounds a lot better than I was imagining
|
|
#
?
Jul 5, 2016 19:27
|
|
|
speaking of OSTree I played around with project atomic while evaluating docker runtime hosts and I really like how it handles system upgrades. good poo poo redhat.
|
|
#
?
Jul 5, 2016 23:36
|
|
|
KDE just released a new version of their desktop. I tried their LiveUSB image. Trip report: it's still bad
|
|
#
?
Jul 6, 2016 00:50
|
|
|
Mr Dog posted:No, although it does kinda seem that way at first glance. i'm not giving access to my home directory to anything directly maintained by gnome upstream those guys are totally fuckso nutballs
|
|
#
?
Jul 6, 2016 01:58
|
|
|
hey remember the loving ridiculous gtk 4.x compatibility plan? https://davmac.wordpress.com/2016/07/05/why-do-we-keep-building-rotten-foundations/ hope you like having literally 20 copies of every library with unknown security holes on your box
|
|
#
?
Jul 6, 2016 01:59
|
|
|
It's a bit of a PR nightmare but there's at least some merit to the argument. Basically, major things have changed in the world of GUIs since Gtk 2 was the latest and greatest. HiDPI emerging on consumer platforms is one, and a universal expectation of tear-free hardware accelerated UI compositing is another. These are breaking changes, you cannot retrofit a repaint-on-expose application to use vsynced double buffering without losing most of the benefit of double buffering. And the Gtk team do actually have some credibility in that they do continue to maintain Gtk 2. Gtk 3 had a bunch of API breakages on account of the whole CSS thing they're doing with widgets. They didn't really have a plan nor a stable release branch that people could target in the meantime. But they are saying that will be stable soon, so Gtk 3 will no longer be such a moving target. BUT, next they want to do this scene graph stuff in order to push more drawing operations into the GPU. Software rendering a UI at 60fps on a 2440x1800 HiDPI display at 60 fps (e.g. when resizing a window) burns more battery and creates more heat than it really ought to. My laptop starts to get uncomfortable if anything uses significant amounts of CPU for an extended period of time. So they're introducing some internal plumbing called GSK, and they're also going to provide it as a scene graph API to application developers directly. People use Clutter for this sort of thing right now, but Gtk+ and Clutter have to do similar things (and mixing them is harder than it should be), so keeping them around as separate projects and continuing to develop them as well results in excessive maintenance burden. So, Clutter and Gtk 3 can be put into maintenance mode (which is a lot less work) and new development can happen in Gtk 4. If your instinct is to complain that Clutter just got pulled out from under you despite the fact that it would continue to receive bug fixes then it sounds like you want change and yet don't want change simultaneously. Provided you believe that they will continue to maintain Gtk 3 while Gtk 4 bakes and then beyond that then this seems to be a decent proposal. They are straight up saying that Gtk 4 will have a whole bunch of churn so you should probably target Gtk 3 for the forseeable. If these revisions are co-installable and continue to be maintained just as Gtk 2 still is then what's the problem? It is possible to manage change in a non-disruptive fashion.
|
|
#
?
Jul 6, 2016 02:19
|
|
|
Notorious b.s.d. posted:hey remember the loving ridiculous gtk 4.x compatibility plan? No, not "literally" 20. Maybe literally two or at most three copies of Gtk itself. You realize Qt4 applications are not exactly extinct in the wild either, right?
|
|
#
?
Jul 6, 2016 02:20
|
|
|
qt4 already supports hi dpi just fine, has done for years and qt5 didn't break every god drat qt4 api none of the gtk3 insanity was really necessary, it's just the worst cadt ever
|
|
#
?
Jul 6, 2016 02:22
|
|
|
Mr Dog posted:No, not "literally" 20. Maybe literally two or at most three copies of Gtk itself. You realize Qt4 applications are not exactly extinct in the wild either, right? every upstream app provider is going to have to ship their own gtk and their own libxml2 and so on because holy lol at the amount of fragmentation we are talking about here if i need gtk 4.6 and openssl 1.0.1f my flatpack isn't going to be able to use your runtime because it has gtk 4.7 or openssl 1.0.1e
|
|
#
?
Jul 6, 2016 02:23
|
|
|
|
| # ? Apr 29, 2024 13:22 |
|
|
Mr Dog posted:If these revisions are co-installable and continue to be maintained just as Gtk 2 still is then what's the problem? It is possible to manage change in a non-disruptive fashion. btw the official plan is to just change the *sonames* not the header names, so they won't be co-installable without shitloads of effort from distributors. you will only be able to build gtk 4.6 applications OR gtk 4.7 applications, not both at once meanwhile, everyone is circulating various plans to remove distributors from the picture this is all going to go wonderful
|
|
#
?
Jul 6, 2016 02:25
|
|