|
api call girl posted:fair enough
|
#
?
Sep 3, 2014 21:42
|
|
|
|
| # ? Oct 14, 2024 06:00 |
|
|
Mr Dog posted:anyway kdbus and gnome sandboxes will solve this for the narrow case of desktop applications (which tbh nobody really cares about for linux anyway). so you'll have a sandboxed GNOME Weather applet and GNOME Music application that can be released to users directly via self-contained release ZIPs from upstream downloaded via the GNOME Software application. Great.
|
|
#
?
Sep 3, 2014 22:17
|
|
|
http://boycottsystemd.org Lmao look at this moron
|
|
#
?
Sep 5, 2014 00:11
|
|
|
api call girl posted:http://boycottsystemd.org yool 1014 monks would have theological debates on how many souls of the damned could have fit in a cask of ale. yool 2014 monks do this.
|
|
#
?
Sep 5, 2014 00:19
|
|
|
if systemd wasn't a thing then i'd say runit is quite nice, it gets the whole orthogonal co-operating processes thing really right. trouble is systemd solves more problems (i.e. actually understanding interprocess dependencies itself instead of expecting an admin or developer to explain them), even if it isn't quite as pretty. i do like pretty things but even i prefer slightly less pretty things that actually get poo poo done. redis is much prettier than postgres and the associated cesspit that is sql, for instance. unfortunately redis requires the db to fit into core and uh can't even filter poo poo when querying a data set, and that's rather important functionality imo. Sapozhnik fucked around with this message at 00:37 on Sep 5, 2014 |
|
#
?
Sep 5, 2014 00:34
|
|
|
Hey instead of running this comprehensive thing I recommend you use these distro-specific bespoke packages that don't even purport to be init replacements Also code xbill to replace bill gates with poettering
|
|
#
?
Sep 5, 2014 00:42
|
|
|
Mr Dog posted:if systemd wasn't a thing then i'd say runit is quite nice, it gets the whole orthogonal co-operating processes thing really right. "lightweight" vs. "solves hard problems"
|
|
#
?
Sep 5, 2014 00:52
|
|
|
Origin posted:yool 1014 monks would have theological debates on how many souls of the damned could have fit in a cask of ale. the "do one thing and do it well" theology coupled with all the world's a file is the worst like, I get the appeal, it was still semi believable 20ish years ago, it's simple and therefore appeals to fresh grads, but it just doesn't stand up to the real world. sometimes you have to violate these principles. Unix did almost from the very beginning, I assume plan 9 tried to be more true but lol plan 9 also lol at the idea that systemd is on the same complexity scale as the kernel itself also if these guys are so desperate for "init reform but MY SIMPLICITY" they should get off their asses and port launchd plus apples log daemon, p sure it does most of the same poo poo without being as complicated. have read poettering's reasons for not just cloning launchd and it was something vague about it not providing certain things enterprise linux wants but hey if you want the simpler version it is out there, and loving well proven irl
|
|
#
?
Sep 5, 2014 00:56
|
|
|
api call girl posted:http://boycottsystemd.org quote:Ultimately, systemd's parasitism is symbolic of something more than systemd itself. It shows a radical shift in thinking by the Linux community. Not necessarily a positive one, either. One that is vehemently postmodern, monolithic, heavily desktop-oriented, choice-limiting, isolationist, reinvents the flat tire, and just a huge anti-pattern in general. If your goal is to pander to the lowest common denominator, so be it. truly the derrida of software for initializing processes
|
|
#
?
Sep 5, 2014 01:02
|
|
|
api call girl posted:http://boycottsystemd.org What about these
|
|
#
?
Sep 5, 2014 01:15
|
|
|
Write your own os/graphics stack/windowing to get away from systemd (and then reimplement xbill as xpoettering)
|
|
#
?
Sep 5, 2014 01:23
|
|
|
BobHoward posted:the "do one thing and do it well" theology coupled with all the world's a file is the worst I seriously wonder if these kinds of assholes were around when BSD became more popular than what AT&T was offering? "Nooooooo I prefer UUCP networking over this TCP crap and gently caress using the Mach VM system MICROKERNALS? a bloo a bloo ugu ~"
|
|
#
?
Sep 5, 2014 01:34
|
|
|
sometimes i think id like to use a tiling wm because i end up tiling everything in normal wms, then i go look at tiling wms and there literally isnt one that doesnt require setting up a config. why would you want some sensible default keybindings to put some windows a screen when you can learn a whole dsl and dependency ecosystem to have a clock
|
|
#
?
Sep 5, 2014 01:39
|
|
|
Origin posted:I seriously wonder if these kinds of assholes were and are grats if you dont have to listen to a coworker complain about linux not being gnu enough on a monthly basis
|
|
#
?
Sep 5, 2014 01:41
|
|
|
Mr Dog posted:app stores don't really work for server applications though, which is why I'm not so keen on this one-size-fits-all solution. There the problem is slightly different: all of the languages used to develop server apps have their own siloed package managers which don't interact with rpm or deb at all (node.js npm, Java maven, Perl CPAN, whatever thing Python has). way back in the day when RPM was being designed, the only one of these siloed package managers that existed was CPAN. RPM knows very well about CPAN and will happily resolve dependencies from CPAN. this seemed like a cool interop feature at the time, but it is never used today the reason is that this turned out to be a horrible mis-feature. filling RPM deps using installed CPAN libraries just makes your configuration unpredictable, because CPAN solves a lot of problems wrong and badly. they all do. rubygems is the worst offender, but cpan and cran and pypi and npm can all gently caress right off for similar reasons for the most part these systems:
RPM does all of these things correctly. Rubygems does none of them. Maven scores 3/5. Pypi scores 1/5. Rate your pet system as you see fit -- i don't spend enough time with npm or cran to really pick them apart.
|
|
#
?
Sep 5, 2014 02:11
|
|
|
it is ok to hate RPM -- it's ugly and painful and riddled with horrible legacy things -- but the things it does properly are not done properly by other things don't talk to me about letting upstream package things until you can show me how my app will reliably depend on signed artifacts (yes, artifacts not source) that i can repeatably deploy maven is far and away the least broken language "silo," but it is still far from perfect (and usually rpm is still your best bet at deploy-time, even if you use maven to build the drat rpm)
|
|
#
?
Sep 5, 2014 02:14
|
|
|
Breakfast All Day posted:sometimes i think id like to use a tiling wm because i end up tiling everything in normal wms, then i go look at tiling wms and there literally isnt one that doesnt require setting up a config. why would you want some sensible default keybindings to put some windows a screen when you can learn a whole dsl and dependency ecosystem to have a clock i3 is p decent by default, but is still probably more useful if you actually learn it. however, like vim, i like it despite having only learned enough to be competent. id kill to have it on windows for work.
|
|
#
?
Sep 5, 2014 02:34
|
|
|
poettering's energies would be better spent developing an rpm and deb replacement to strongarm everyone into switching to and have its installation processing be declarative with a fixed vocabulary that can maybe be extended over time. deb went the opposite way: debs used to have completely freeform installation shell scripts that ran as root but over time bits and pieces of these all got packaged into "debhelper" scripts and now most debs install themselves exclusively by calling a sequence of debhelper scripts from the installation script and nothing else or something. i've never actually built a deb. make shared library developer ppl do some semver type thing and have all these different .so builds install side-by-side. use debian's multiarch approach to handle x86 and x86_64 on the same system because that is the Correct Solution to this problem. nobody has ever explained to me why we need crazy filesystem namespacing bullshit to install /usr/lib/libdickbutt.so.1.5.3 and /usr/lib/libdickbutt.1.6.2 side-by-side (and have .1.5.3 automatically replace 1.5.2) with the package manager maintaining its own index of the available .so versions installed on the system and available in the package repository. u can then install a package produced by upstream and if like, it says it needs shared libs X Y and Z and DBus services org.foo and org.bar then whatever distro ur running the pkg manager should be able to satisfy them. maybe even have some URLs where upstream (signed!) versions can be downloaded (and where security-patched updated binaries can also be updated as they become available) then go through the absolutely monumental effort of making everything produce deterministic builds that have a cryptographically secure link to the Git commit used to produce each of these binaries and i'm already gettin a semi thinking about it or, you know, do a lovely half-assed solution where you stuff a bunch of binaries that came from gently caress-knows-where into a gigantic loving tarball and sucks to be you if /usr/@SOMERUNTIMEPACKAGE-JUSTGONNASHITTHISALLOVERYOURFILESYSTEM-1.27.3.00176543-f00cface/lib64/libz.so turns out to contain a buffer overflow
|
|
#
?
Sep 5, 2014 02:35
|
|
|
Mr Dog posted:poettering's energies would be better spent developing an rpm and deb replacement to strongarm everyone into switching to have u considered...isntalling gentoo?
|
|
#
?
Sep 5, 2014 02:46
|
|
|
The issue with making a better deb/rpm packages is that at the end of the day... they're still deb/rpm. deb/rpm have three things: the file format itself, the user runtime, and the build system and infrastructure around them. As a developer, the build system and infrastructure around them is terrible: I don't like duplicating all my commit messages into four different changelogs (upstream git log, debian changelog/rpm specfile, package git log, RHEL-mandated CHANGELOG file). I don't like the 80s-style workflow of patch files in a directory, rather than just building from a git branch. My workflow should be "point an automated tool at this git branch". As a user, I'm not a fan of the user runtime, either. I don't like how libpng means that I can't upgrade Firefox, or if Firefox fails to upgrade, I can't install an upgrade to LibreOffice separately. Bundling system components and applications together in the same namespace has always provided for a terrible user experience. Unlike Poettering's solution, I've developed an app bundle system that *can* deal with security updates, applicable by the system admin, not the runtime developer. I think it's inexcusable to not have a way for the sysadmin to hotpatch your system in a sane way, especially with how 0days develop in this climate. And then there's the file format, which is what all the arguments are about, and that I don't give a poo poo about. Metadata's metadata, at the end of the day.
|
|
#
?
Sep 5, 2014 02:52
|
|
|
Yet these people still don't want to use FreeBSD, OpenBSD, NetBSD, or DragonBSD.
|
|
#
?
Sep 5, 2014 03:37
|
|
|
MrMoo posted:Yet these people still don't want to use FreeBSD, OpenBSD, NetBSD, or DragonBSD. tell your boss you want to go to production on openbsd, report back
|
|
#
?
Sep 5, 2014 03:48
|
|
|
the reason systemd is controversial is that it has been picked up by redhat, the 800 lb gorilla. there is no meaningful "choice" there.
|
|
#
?
Sep 5, 2014 03:48
|
|
|
so is lunix available on the desktop yet itt
|
|
#
?
Sep 5, 2014 03:53
|
|
|
theadder posted:so is lunix available on the desktop yet itt Yep. We've sold all 100 of our prototype computers to real users. Unlike most valley companies, we actually have revenue.
|
|
#
?
Sep 5, 2014 04:32
|
|
|
theadder posted:so is lunix available on the desktop yet itt 
|
|
#
?
Sep 5, 2014 04:34
|
|
|
i am microtel
|
|
#
?
Sep 5, 2014 04:42
|
|
|
http://www.microtelpc.com/ still a website check out they pcs
|
|
#
?
Sep 5, 2014 04:42
|
|
|
MCS6013 • Intel Core i7 920 • 3GB DDR3-1333 memory • NVIDIA 250 GTS 1GB Video • 24x DVD+-RW Drive From $899 
|
|
#
?
Sep 5, 2014 04:55
|
|
|
Notorious b.s.d. posted:tell your boss you want to go to production on openbsd, report back seems like it would be OK as a firewall or route server
|
|
#
?
Sep 5, 2014 06:05
|
|
|
We use OpenBSD + OpenVPN for remote access to clients sites. Just searching the company document share and it looks like we even have production systems written in Go on OpenBSD 
|
|
#
?
Sep 5, 2014 06:18
|
|
|
any vision of linux succeeding on the desktop that somehow involves libreoffice is ridiculous on its face gdocs is a more viable route since it actually does some things better than the thing people already have (cost is an illusion, people get office), but that p. means linux on the desktop would be chromeos which community-wise is p. much worse than microsoft staying on top since more foss stuff happens on windows than on chromeos
|
|
#
?
Sep 5, 2014 06:39
|
|
|
it has no monitor because the graphics drivers don't work.
|
|
#
?
Sep 5, 2014 11:45
|
|
|
computer toucher posted:it has no monitor because the graphics drivers don't work. But it has speakers? Something is rotten in the state of Denmark
|
|
#
?
Sep 5, 2014 13:13
|
|
|
Origin posted:"Nooooooo I prefer UUCP networking over this TCP crap and gently caress using the Mach VM system MICROKERNALS? a bloo a bloo ugu ~" at least the latter was something a lot of people actually whined and complained about back in the Net2/386BSD days. those of us with some exposure to Mach knew better
|
|
#
?
Sep 6, 2014 04:04
|
|
|
Breakfast All Day posted:grats if you dont have to listen to a coworker complain about linux not being gnu enough on a monthly basis this is something I never hear at my job and I'm very happy for it
|
|
#
?
Sep 6, 2014 04:05
|
|
|
Suspicious Dish posted:sold what
|
|
#
?
Sep 6, 2014 04:18
|
|
|
eschaton posted:this is something I never hear at my job and I'm very happy for it
|
|
#
?
Sep 6, 2014 04:20
|
|
|
eschaton posted:what We're a small company, we can't support ten thousand users overnight. We made 100 of the bamboo machines as a test run off our line, 50 of them shipped to Guatemala, and they were sold to real users who were aware this was a beta. We're going to be doing user testing and infrastructure rollout on these 50. All that will change between this and the full 100,000 first production line is the image shipped on them.
|
|
#
?
Sep 6, 2014 04:34
|
|
|
|
| # ? Oct 14, 2024 06:00 |
|
|
Notorious b.s.d. posted:the reason systemd is controversial is that it has been picked up by redhat, the 800 lb gorilla. there is no meaningful "choice" there.
|
|
#
?
Sep 6, 2014 04:41
|
|