|
https://www.kde.org/info/security/advisory-20180208-2.txt nice
|
#
?
Feb 13, 2018 19:58
|
|
|
|
| # ? Apr 25, 2024 13:09 |
|
|
ratbert90 posted:How about we laugh at AMD? the problem wasn't the size of the code; gpus and display handling are really complicated and need a lot of backend driver code to work. the problem was that the original code was opened by throwing it over the fence and had a lot of dead and unused bits, multiple reimplementations of systems the kernel already provides and lots of strange stylistic things like storing initialization values for registers as strings of binary that were parsed at run time instead of just, you know, putting the constant values directly in the code.
|
|
#
?
Feb 13, 2018 22:46
|
|
|
Sapozhnik posted:i mean if you want to poo poo something proprietary with its own private SOs into a subdir of /opt nothing is stopping you? chrome and other such things install themselves there chrome is an example of a serious offender, twice over:
major bummers. party foul, if you will Sapozhnik posted:i don't get what point you are trying to make; a distribution uses a frozen slice of shared library versions, that's what a distribution is. if you want a hermetic environment for your programs that's what flatpak and (ugh) docker are for. or RPMs that install to /opt. Suspicious Dish posted:yeah, i'm really happy that linux solved dll he-- /usr/lib/libstdc++.so.6: version `GLIBCXX_3.4.9' not found suspicious dish pretty much nailed the problem even inside my "frozen slice" of shard library versions, I am very likely to have multiple versions (chronologically) or multiple ABIs (what if my binary is more than one year old?) or multiple versions (differing vendors) those need to be stored in separate paths in order to avoid dumbassery. the runtime linker should examine only the correct paths, not just load everything in /usr/lib and cross our fingers again: svr4 did this correctly, with full package management, natch, in nineteen eighty eight how is it 2018 and a major software vendor is shipping shell scripts with LD_LIBRARY_PATH
|
|
#
?
Feb 14, 2018 06:43
|
|
|
we learn nothing solaris 2.0 did all of this poo poo perfectly 28 years ago. the necessary specifications and reference implementations were published 30 years ago. now it is 2018 and we are arguing over several mutually-incompatible ways to do completely the wrong loving thing
|
|
#
?
Feb 14, 2018 06:45
|
|
|
too bad for the solaris people that scott mcnealy was running the company and his business strategy was seeing how many times he could call bill gates a poopyhead
|
|
#
?
Feb 14, 2018 06:50
|
|
|
rpath doesn’t help when multiple versions of the same library get loaded into a process and the second loser gets a mishmash of symbol versions because ELF lookup semantics are idiotic
|
|
#
?
Feb 14, 2018 07:25
|
|
|
Notorious b.s.d. posted:chrome is an example of a serious offender, twice over: how the hell are you not able to solve this by having your own lib folder with your binary in /opt. i still dont see what your problem is, it seems like you are hellbent on creating them
|
|
#
?
Feb 14, 2018 16:55
|
|
|
Tankakern posted:how the hell are you not able to solve this by having your own lib folder with your binary in /opt. i still dont see what your problem is, it seems like you are hellbent on creating them having your own lib folder with your binary and setting rpath is the correct solution, yes everything on the system should be doing this, not just a handful of sane people who care whether their poo poo works i am not, for example, going to ship my own libstdc++.so.6, for obvious reasons. but i can still run into versioning issues because the distributor just dumps a single copy in /usr/lib and calls it done
|
|
#
?
Feb 14, 2018 18:54
|
|
|
Notorious b.s.d. posted:i am not, for example, going to ship my own libstdc++.so.6, for obvious reasons. but i can still run into versioning issues because the distributor just dumps a single copy in /usr/lib and calls it done …and doesn’t care if next year’s OS exports the same API with the same behavior from a lib in the same place with the same name if it’s public API it needs to be binary compatible if it’s not going to be binary compatible it needs to not be public API this is not a hard concept to grasp but Linux has always been “just recompile, not having the sources necessary to do so is a moral failing” never mind that even if all the sources are available you also have to re-test everything
|
|
#
?
Feb 15, 2018 08:46
|
|
|
eschaton posted:just recompile, not having the sources necessary to do so is a moral failing unironically 
|
|
#
?
Feb 15, 2018 09:03
|
|
|
eschaton posted:Linux... never mind that... you also have to test
|
|
#
?
Feb 15, 2018 09:26
|
|
|
Gazpacho posted:too bad for the solaris people that scott mcnealy was running the company and his business strategy was seeing how many times he could call bill gates a poopyhead also hosting hockey games
|
|
#
?
Feb 15, 2018 09:49
|
|
|
i don't know why this made me laugh i'm sorry
|
|
#
?
Feb 15, 2018 18:51
|
|
|
it wasn't that funny and indirectly lead me to https://www.reddit.com/r/freebsd/comments/7xapx2/freebsds_new_geek_feminismbased_code_of_conduct/ and now i am angry
|
|
#
?
Feb 15, 2018 19:06
|
|
|
the pissing and moaning about systemd is never going to go away is it
|
|
#
?
Feb 15, 2018 19:49
|
|
|
Sapozhnik posted:the pissing and moaning about systemd is never going to go away is it no, i meet new and new people who use gentoo at home and become 300 wpm typists when i mention systemd
|
|
#
?
Feb 15, 2018 19:51
|
|
|
of course, grudges about stupid poo poo are forever, people still bitch about minutiae from decades ago
|
|
#
?
Feb 15, 2018 19:51
|
|
|
Tankakern posted:the dll hell is still there, it's just abstracted behind winsxs, bloating the os to new extremes, making every version of every dll released ever available how is this not a correct solution
|
|
#
?
Feb 15, 2018 19:57
|
|
|
winsxsaas
|
|
#
?
Feb 15, 2018 19:57
|
|
|
I tried playing around with MSYS2 for Windows development recently. It uses Pacman instead of cygwin's weird setup.exe thing, which is nice. What isn't nice is the fact that installs separate MinGW-targeting MSYS2 environments that you have to open up as separate shells instead of just installing them as cross toolchains like every other Linux-based toolchain install. Swear to god nobody can get this poo poo right ever. I'd use WSL but that involves Windows 10 and at this point I'm only going to do that if I can somehow pirate the Enterprise Edition. I'm not going to use MSVC because it still doesn't loving support C99, a standard that's almost old enough to drink
|
|
#
?
Feb 15, 2018 20:08
|
|
|
Nobody even explains this crap either: Cygwin, the library: A fake libc that provides fake POSIX syscalls under Windows. Also redirect filesystem syscalls to be rooted under a UNIXy filesystem in C:\cygwin64 or whatever, though you can access the entire win32 filesystem under /cygwin/c/ or whatever. Cygwin, the distribution: A UNIX software distribution compiled against Cygwin, the DLL, using a GCC tool chain targetting the Cygwin library. MinGW: A GCC target that targets native the Win32 API without the Cygwin library being involved. Also includes w32headers, which is a set of permissively-licensed header files for the Win32 API. This toolchain can be installed under Cygwin (the distribution), MSYS2 (see below), or on Linux or macOS. If installed on a Windows host it does require Cygwin the library for the compiler to function, but its outputs do not require the Cygwin library. MinGW-w64: A fork of MinGW and w32headers that isn't moribund. Despite the name this supports both 32-bit and 64-bit Windows, it was merely created to facilitate 64-bit Windows support since the original project never got around to it. MSYS2: An alternative to Cygwin, the distribution. But it also uses Cygwin, the library, which may or may not also be forked idk. Main difference is that it uses Arch's pacman package manager instead of having to go through a lovely native win32 wizard every time you want to install a package. Also the aforementioned separate-installs-for-each-toolchain BS. And symlinks being implemented as deep copies by default unless you edit this weird semi-undocumented flag somewhere.
|
|
#
?
Feb 15, 2018 20:14
|
|
|
Cybernetic Vermin posted:it wasn't that funny and indirectly lead me to https://www.reddit.com/r/freebsd/comments/7xapx2/freebsds_new_geek_feminismbased_code_of_conduct/ and now i am angry tech_poisonous_environment.epub some jackass posted:
 as above posted:Linux is much worse for this sort of thing. The Kernel is safe and so are old guard distros. But giant swathes of the software has been compromised. Not that I would use it anyway, but Gnome is one of the worst offenders. didn't know that anti-harassment policies could "compromise" a desktop environment but the best for last: wat posted:
Best Bi Geek Squid fucked around with this message at 20:30 on Feb 15, 2018 |
|
#
?
Feb 15, 2018 20:26
|
|
|
Sapozhnik posted:I tried playing around with MSYS2 for Windows development recently. It uses Pacman instead of cygwin's weird setup.exe thing, which is nice. What isn't nice is the fact that installs separate MinGW-targeting MSYS2 environments that you have to open up as separate shells instead of just installing them as cross toolchains like every other Linux-based toolchain install. you deffo can pirate windows 10 enterprise, but it may involve using russian cracks
|
|
#
?
Feb 15, 2018 20:28
|
|
|
Sapozhnik posted:I'm not going to use MSVC because it still doesn't loving support C99, a standard that's almost old enough to drink i thought they actually implemented a fair amount of c99 as of the 2013 version, at least enough to compile ffmpeg
|
|
#
?
Feb 15, 2018 21:01
|
|
|
cinci zoo sniper posted:you deffo can pirate windows 10 enterprise, but it may involve using russian cracks isn't there a reverse engineered kms server out there?
|
|
#
?
Feb 15, 2018 21:13
|
|
|
dealing with windows 10 makes me want to kms
|
|
#
?
Feb 15, 2018 21:14
|
|
|
SamDabbers posted:isn't there a reverse engineered kms server out there? not sure, i just know that some of my acquaintances use enterprise lts skus at home, and most of them are unemployed
|
|
#
?
Feb 15, 2018 21:15
|
|
|
Sapozhnik posted:dealing with windows 10 makes me want to kms its the best os ive used
|
|
#
?
Feb 15, 2018 21:15
|
|
|
cinci zoo sniper posted:its the best os ive used i'm sorry
|
|
#
?
Feb 15, 2018 21:16
|
|
|
Sapozhnik posted:I'm not going to use MSVC because it still doesn't loving support C99, a standard that's almost old enough to drink msvc supports C99 to the letter, but for maximum comedy value it only applies if you're writing actual c in a file named .c
|
|
#
?
Feb 15, 2018 21:23
|
|
|
oh. well maybe i should take a look at it then does it have stdatomic.h cuz that's the only c11 feature i care about at this time
|
|
#
?
Feb 15, 2018 21:25
|
|
|
Sapozhnik posted:dealing with windows 10 makes me want to kms
|
|
#
?
Feb 15, 2018 21:43
|
|
|
Sapozhnik posted:oh. well maybe i should take a look at it then nah, they only pulled C99 in because most of it is required for C++11
|
|
#
?
Feb 15, 2018 22:37
|
|
|
Sapozhnik posted:dealing with windows 10 makes me want to kms
|
|
#
?
Feb 15, 2018 22:50
|
|
|
cinci zoo sniper posted:no, i meet new and new people who use gentoo at home and become 300 wpm typists when i mention systemd said this before many times, but i hate the fact that gentoo is so closely associated with anti-systemd retards. gentoo with systemd is the best distro there is, they just so happen to also support a not-systemd init system
|
|
#
?
Feb 15, 2018 23:02
|
|
|
Tankakern posted:retards. out.
|
|
#
?
Feb 15, 2018 23:03
|
|
|
Tankakern posted:
|
|
#
?
Feb 15, 2018 23:20
|
|
|
Tankakern posted:said this before many times, but i hate the fact that gentoo is so closely associated with anti-systemd retards. gentoo is associated with every kind of mouth-breather there is. fedoras, beards, systemd weirdos, cranks you name it it's almost like an ill-conceived bundle of idiot patchsets only attracts morons -- "what if we had an os that was literally impossible to patch faster than it became exploitable?" -- the creed of gentoo, 1999
|
|
#
?
Feb 15, 2018 23:22
|
|
|
Chrome OS is a Gentoo fork.
|
|
#
?
Feb 15, 2018 23:31
|
|
|
|
| # ? Apr 25, 2024 13:09 |
|
|
Most successful Linux Desktop distro.
|
|
#
?
Feb 15, 2018 23:31
|
|