|
Visions of Valerie posted:good news, the C ABI already exists on your OS and if you're not targeting it you should be another thing windows gets right is making unwinding part of the platform ABI and presenting segfaults to the process as exceptions
|
#
?
Feb 21, 2024 05:36
|
|
|
|
| # ? Jun 11, 2024 22:14 |
|
|
Subjunctive posted:making an ABI without a programming language in mind seems like it’ll result in a system that sucks to use from all languages, instead of all-but-one languages drat didnt have you pegged as a nosql liker. learn something every day edit: im just saying lets really put that "computers are fast enough now" idea to the test and make stack frames out of json. free named params. human readable core dumps. its a no brainer Nomnom Cookie fucked around with this message at 05:52 on Feb 21, 2024 |
|
#
?
Feb 21, 2024 05:45
|
|
|
stretch goal: use jq to implement a CPS interpreter
|
|
#
?
Feb 21, 2024 05:55
|
|
|
abi word
|
|
#
?
Feb 21, 2024 06:15
|
|
|
blinkenlights on flippy switch vacuum tube monstrosity
|
|
#
?
Feb 21, 2024 06:19
|
|
|
Nomnom Cookie posted:another thing windows gets right is making unwinding part of the platform ABI and presenting segfaults to the process as exceptions I don't hate the signal(7) interface but I also rarely want to do anything on SEGV other than dump core so 
|
|
#
?
Feb 21, 2024 07:22
|
|
|
You should, there's no way to use them from multiple threads or different parts of the program.
|
|
#
?
Feb 21, 2024 08:01
|
|
|
Subjunctive posted:depending so tightly on implementation characteristics behind IWhatever that you can’t actually substitute anything anyway so just like every other interface
|
|
#
?
Feb 21, 2024 08:40
|
|
|
pseudorandom name posted:dynamic linking works great on operating systems designed by professionals yeah, it'll be great if we ever get one someday
|
|
#
?
Feb 21, 2024 08:41
|
|
|
likely would have been better had dynamic linking been confined to a well-defined set of platform libraries a vendor is standing prepared to maintain forward compatibility for. the idea of more broadly randomly swapping out large chunks of programs from under them never worked and never will.
|
|
#
?
Feb 21, 2024 10:20
|
|
|
Cybernetic Vermin posted:likely would have been better had dynamic linking been confined to a well-defined set of platform libraries a vendor is standing prepared to maintain forward compatibility for. the idea of more broadly randomly swapping out large chunks of programs from under them never worked and never will. Is this not how it worked on SunOS originally?
|
|
#
?
Feb 21, 2024 12:05
|
|
|
Cybernetic Vermin posted:likely would have been better had dynamic linking been confined to a well-defined set of platform libraries a vendor is standing prepared to maintain forward compatibility for. the idea of more broadly randomly swapping out large chunks of programs from under them never worked and never will. sure is a good thing for this argument that Linux distros don't exist
|
|
#
?
Feb 21, 2024 17:40
|
|
|
Visions of Valerie posted:sure is a good thing for this argument that Linux distros don't exist rather key that they exist as the cautionary example
|
|
#
?
Feb 21, 2024 17:43
|
|
|
Cybernetic Vermin posted:rather key that they exist as the cautionary example idk mate "Debian/Ubuntu/Fedora/etc. never worked and never will" is a pretty galaxy brain take
|
|
#
?
Feb 21, 2024 17:51
|
|
|
had you been clever you would have dropped in rhel (or if a certain kind of sicko argument nix) but lol ubuntu working
|
|
#
?
Feb 21, 2024 17:58
|
|
|
Cybernetic Vermin posted:had you been clever you would have dropped in rhel (or if a certain kind of sicko argument nix) but lol ubuntu working have you been paying attention to red hat lately
|
|
#
?
Feb 21, 2024 18:08
|
|
|
Visions of Valerie posted:have you been paying attention to rhel lately RHEL 9 is fine.
|
|
#
?
Feb 21, 2024 18:09
|
|
|
FlapYoJacks posted:RHEL 9 is fine. centos sure ain't though
|
|
#
?
Feb 21, 2024 18:09
|
|
|
Visions of Valerie posted:centos sure ain't though Absolutely. Don’t use CentOS. I think Rocky is the new CentOS.
|
|
#
?
Feb 21, 2024 18:10
|
|
|
rhel being an ibm product brings some worries, but among them are *not* matters of support, updates, and forward compatibility. if you can pay.
|
|
#
?
Feb 21, 2024 18:25
|
|
|
FlapYoJacks posted:Absolutely. Don’t use CentOS. I think Rocky is the new CentOS. rocky is the new centos but a lot of vendor garbage that ran on centos doesn't run on rocky because parsing /etc/redhat-release is hard
|
|
#
?
Feb 21, 2024 19:50
|
|
|
Subjunctive posted:runtime symbol resolution for me, but not for thee i like what sdl does, if you static link they slot in a jump table anyway. so nowadays my feelin is: why not both
|
|
#
?
Feb 21, 2024 20:55
|
|
|
Visions of Valerie posted:sure is a good thing for this argument that Linux distros don't exist do you really look at a thousand nerds pushing on dependency conflicts until they hack together a pile that appears to work and think "yes, this is the Absolute Idea of operating system design"
|
|
#
?
Feb 21, 2024 21:40
|
|
|
thats all os's, you're just picking between nerds gettin paid figs at the actual os company and an alliance of nerds gettin paid figs at the places that wanna screw the os company over and true ideologues
|
|
#
?
Feb 21, 2024 21:43
|
|
|
Linux distros used to break my product all the time by trying to be clever and split it up into more shared libraries (hurting performance along the way). grr
|
|
#
?
Feb 21, 2024 21:45
|
|
|
Cybernetic Vermin posted:likely would have been better had dynamic linking been confined to a well-defined set of platform libraries a vendor is standing prepared to maintain forward compatibility for. the idea of more broadly randomly swapping out large chunks of programs from under them never worked and never will. like Windows
|
|
#
?
Feb 21, 2024 22:24
|
|
|
you could also build a application patching system into the OS itself, for backwards compatibility with particularly broken applications like Windows
|
|
#
?
Feb 21, 2024 22:25
|
|
|
i guess the mac peeps have things a fair tad better
|
|
#
?
Feb 21, 2024 22:26
|
|
|
no ebpf or WPF, so that’s unfortunate. don’t think you can drive the profiler from an API any more to control the workload that you want to focus on; it was really handy to be able to start/stop/pause/resume to get cleaner profiles
|
|
#
?
Feb 21, 2024 22:40
|
|
|
bob dobbs is dead posted:i guess the mac peeps have things a fair tad better no muss no fuss
|
|
#
?
Feb 21, 2024 22:46
|
|
|
my linux distro story is that i discovered that one of my project's dependencies was commonly packaged in a way that didn't work for my project and would result in a crash on startup. i added a configure check to validate that the dependency was configured in a compatible way and would fall back to building it from source if not. multiple package maintainers just patched this check out, resulting in the application crashing on startup.
|
|
#
?
Feb 21, 2024 22:50
|
|
|
I wonder if they’ll open source Rosetta 2 after they don’t need it anymore. well, I’m sure they won’t because that’s not how they roll, but it would be nice! wouldn’t really help their real competitors because those chips don’t have the extensions for running the x86 memory model, but I bet the qemu kids could get some mileage out of it, and maybe the game emulators!
|
|
#
?
Feb 21, 2024 22:51
|
|
|
Slackware on floppy disks was my first distro. 
|
|
#
?
Feb 21, 2024 22:52
|
|
|
Subjunctive posted:I wonder if they’ll open source Rosetta 2 after they don’t need it anymore. well, I’m sure they won’t because that’s not how they roll, but it would be nice! wouldn’t really help their real competitors because those chips don’t have the extensions for running the x86 memory model, but I bet the qemu kids could get some mileage out of it, and maybe the game emulators! it'd definitely be neat to see under the hood of rosetta 2 and figure out what makes it tick, but I imagine that the code is pretty grody
|
|
#
?
Feb 21, 2024 22:54
|
|
|
FlapYoJacks posted:Slackware on floppy disks was my first distro. on UMSDOS with LOADLIN?
|
|
#
?
Feb 21, 2024 22:56
|
|
|
pseudorandom name posted:on UMSDOS with LOADLIN? Ya goddamn straight. I switched to ext2 shortly after and lilo.
|
|
#
?
Feb 21, 2024 22:57
|
|
|
its time for your colonoscopy
|
|
#
?
Feb 21, 2024 22:58
|
|
|
pseudorandom name posted:its time for your colonoscopy literally 2 years away lol.
|
|
#
?
Feb 21, 2024 22:58
|
|
|
Kazinsal posted:it'd definitely be neat to see under the hood of rosetta 2 and figure out what makes it tick, but I imagine that the code is pretty grody all code is grody when you understand it FlapYoJacks posted:Slackware on floppy disks was my first distro. HJ Lu's boot/root? then something that didn't like something about my motherboard, then SLS for a while
|
|
#
?
Feb 21, 2024 23:40
|
|
|
|
| # ? Jun 11, 2024 22:14 |
|
|
Plorkyeran posted:my linux distro story is that i discovered that one of my project's dependencies was commonly packaged in a way that didn't work for my project and would result in a crash on startup. i added a configure check to validate that the dependency was configured in a compatible way and would fall back to building it from source if not. multiple package maintainers just patched this check out, resulting in the application crashing on startup. hey, at least it crashed after the distro patched it. instead of, idk, silently producing secret keys suitable for use in a TSA lock. but hey at least debian didn't have to rebuild the entire archive to fix their fuckup!
|
|
#
?
Feb 22, 2024 00:58
|
|