|
headshot
|
#
?
Jan 31, 2016 01:31
|
|
|
|
| # ? Apr 20, 2024 02:58 |
|
|
eschaton posted:I remember when you'd get a box of 3.5in disks with the 386BSD system on them yeah its weird. i was wondering the other day about why linux took off even though bsd had been around since the late 70's. maybe some weird legal thing? the folk-story in my head about linux was that it was "free", where as the bsd source tree floating around had some legal miasma around it.
|
|
#
?
Jan 31, 2016 03:29
|
|
|
http://www.midnightbsd.org/screenshots/
|
|
#
?
Jan 31, 2016 04:26
|
|
|
there are tons of bsd 'distros' fyi
|
|
#
?
Jan 31, 2016 04:27
|
|
|
pram posted:there are tons of bsd 'distros' fyi misread that as "bad" and couldn't disagree
|
|
#
?
Jan 31, 2016 04:37
|
|
|
I'm the "HTML5 Lightbox Free Version" text in the Lightbox/color box modal when you click a screenshot to make it bigger craisins fucked around with this message at 06:21 on Jan 31, 2016 |
|
#
?
Jan 31, 2016 04:40
|
|
|
craisins posted:I'm the "HTML5 Lightbox Free Version" text in the Lightbox/color box modal when you click a screenshot to make it bigger n free as in freedom
|
|
#
?
Jan 31, 2016 04:45
|
|
|
Barnyard Protein posted:yeah its weird. i was wondering the other day about why linux took off even though bsd had been around since the late 70's. maybe some weird legal thing? the folk-story in my head about linux was that it was "free", where as the bsd source tree floating around had some legal miasma around it. linux landed a year earlier than the x86 bsd versions and bsd was initially tied up in a lawsuit between at&t and uc berkley over ownership of the code. the lawsuit wasn't settled until 1994.
|
|
#
?
Jan 31, 2016 04:53
|
|
|
i think freebsd has a single benefit over linux, and thats the kernel and userland are maintained and developed by the same people. other than that its an anachronism and i dont see the point
|
|
#
?
Jan 31, 2016 04:57
|
|
|
also the license means you can turn it into over things. like the ps4 did.
|
|
#
?
Jan 31, 2016 05:57
|
|
|
The_Franz posted:linux landed a year earlier than the x86 bsd versions and bsd was initially tied up in a lawsuit between at&t and uc berkley over ownership of the code. the lawsuit wasn't settled until 1994. the article series about 386BSD started in early 1991
|
|
#
?
Jan 31, 2016 06:52
|
|
|
Barnyard Protein posted:yeah its weird. i was wondering the other day about why linux took off even though bsd had been around since the late 70's. maybe some weird legal thing? the folk-story in my head about linux was that it was "free", where as the bsd source tree floating around had some legal miasma around it. The_Franz posted:linux landed a year earlier than the x86 bsd versions and bsd was initially tied up in a lawsuit between at&t and uc berkley over ownership of the code. the lawsuit wasn't settled until 1994. yeh there was very real concern that bsd wasnt going to survive the lawsuit. fear lingered even after the lawsuit because even though bsd wasnt killed the uc regents and at&t chose to do a settlement with secret terms, which left a lot of open questions. i remember seeing usenet discussions about how linux was a ray of hope for open source unix because as a clean sheet kernel it wasn't linked to at&t unix intellectual property in any way so who could sue??? ironically, that turned out not to be a shield against getting sued, though sco's... novell (ahem) legal theories didn't work out in court the other factor (or so i've heard it) was that linus torvalds was considerably more open to outside input than the overlords of bsd, which attracted lots of new blood. at the time people jumped on board there certainly werent any technical merits to linux over a bsd, early linux was exactly the sort of hilarious halfassed shitpile you'd expect of some random finnish kid's literal hobby project os. another contestant was minix but it took itself out of the running by tanenbaum insisting the code remain simple enough to use as course material in undergrad operating systems internals classes, so it was kind of a toy
|
|
#
?
Jan 31, 2016 07:10
|
|
|
Qnx still going strong too
|
|
#
?
Jan 31, 2016 09:15
|
|
|
a pretty useful toy though, especially as people added support for large memory, long filenames, and even networking (which I think required a new server atop the microkernel, peer to the filesystem and memory manager) I added the most basic possible 68040 support myself, to flush the instruction & data caches on context switch so '040 systems could run with caches enabled I need to find my boxed copy of MacMinix so I can extract a pristine version of the Minix 1.5.10 sources for historically-inclined folks, there doesn't seem to be any other clean copy around
|
|
#
?
Jan 31, 2016 09:18
|
|
|
Suspicious Dish posted:uh there are lots of valid cases to writing to the system efi partition like changing boot configuration? The EFI system partition is where the bootloader lives, not your firmware's boot configuration. Writeable EFI System Partition is fine. Writeable loving filesystem metaphor for system configuration nvram is a hideously stupid loving idea. Both the filesystem metaphor part and the mounting it writeable part. Crowbarring system status and configuration APIs into a filesystem syscall interface is the stupidest goddamn poo poo. Sapozhnik fucked around with this message at 17:21 on Jan 31, 2016 |
|
#
?
Jan 31, 2016 17:15
|
|
|
linux really overtook bsd when it came to smp. they made the leap to fine-grained multithreaded kernel earlier while bsd was stuck with the big kernel lock for years.
|
|
#
?
Jan 31, 2016 17:23
|
|
|
Linux prior to at the very least 2.6.0 was a toy. I'm not entirely sure where exactly the momentum came from but I'm guessing it was the GPL forcing everybody to play nice, so big vendors like IBM could do very commercially valuable work like NPTL (non-toy multithreading, which involved both userspace and kernelspace work) and not have their competitors turn around and go "lol thanks for the free R&D" and immediately stuff it into their OS product. The BSD and commercial UNIX landscape was balkanized and tragedy of the commons'ed to hell because of the BSD license that their libertarian sperglord community loves so much. e: my bad, IBM did the competing NGPT project, NPTL was Red Hat
|
|
#
?
Jan 31, 2016 17:28
|
|
|
pram posted:i think freebsd has a single benefit over linux, and thats the kernel and userland are maintained and developed by the same people. other than that its an anachronism and i dont see the point that can also be a hindrance since the centralized development of the linux kernel means that things like device drivers only need to concern themselves with "the kernel" to be supported on every distro. aside from the repacks of freebsd, every bsd fork is a separate special snowflake project which may occasionally share code but are otherwise independent and need to be dealt with separately.
|
|
#
?
Jan 31, 2016 17:30
|
|
|
I'd say 2.4 was the first non-garbage kernel, made better if you used the redhat nptl backport.
|
|
#
?
Jan 31, 2016 17:41
|
|
|
I remember udev being a big deal and modifying the nvidia driver installer to work on kernel 2.5. 
|
|
#
?
Jan 31, 2016 18:16
|
|
|
The Management posted:linux really overtook bsd when it came to smp. they made the leap to fine-grained multithreaded kernel earlier while bsd was stuck with the big kernel lock for years. freebsd fixed smp before linux.
|
|
#
?
Jan 31, 2016 18:17
|
|
|
The Management posted:I'd say 2.4 was the first non-garbage kernel, made better if you used the redhat nptl backport. 2.4 was a flaming shitbag disaster. literally 5 years of crashes and buggy drivers. they unironically called it the kernel of pain. that was the biggest presentation at linuxworld 2002.
|
|
#
?
Jan 31, 2016 18:18
|
|
|
BobHoward posted:the other factor (or so i've heard it) was that linus torvalds was considerably more open to outside input than the overlords of bsd, which attracted lots of new blood. it's true that bill and lynne jolitz were not very responsive to community beardos submitting patches. that's where netbsd and freebsd came from. it's not that netbsd and freebsd dind't accept patches. they were founded specifically for that purpose
|
|
#
?
Jan 31, 2016 18:22
|
|
|
i can't believe you guys have made me defend bsd. pram is right, bsd, all of them, are anachronisms. but getting the history wrong makes me crazy. there was a time when freebsd and netbsd were relevant and important in the industry. those who do not understand history are doomed to repeat it etc
|
|
#
?
Jan 31, 2016 18:23
|
|
|
Mr Dog posted:Crowbarring system status and configuration APIs into a filesystem syscall interface is the stupidest goddamn poo poo. it wouldn't be so bad if there were proper system call apis to accompany them. having to traverse a directory tree just to know how many processors are in the machine you are running on or needing to parse a bunch of text from /proc/meminfo to find out the system memory stats is loving stupid and a pain in the rear end
|
|
#
?
Jan 31, 2016 18:33
|
|
|
i just deleted my efi folder on /sys/ firmware/efi and my motherboard regenerated it on reboot, so i guess it depends on your motherboard.
|
|
#
?
Jan 31, 2016 18:55
|
|
|
it just works
|
|
#
?
Jan 31, 2016 19:00
|
|
|
The_Franz posted:it wouldn't be so bad if there were proper system call apis to accompany them. having to traverse a directory tree just to know how many processors are in the machine you are running on or needing to parse a bunch of text from /proc/meminfo to find out the system memory stats is loving stupid and a pain in the rear end does Linux still not have a standard dictionary object format that can return complex data?
|
|
#
?
Jan 31, 2016 19:08
|
|
|
The_Franz posted:it wouldn't be so bad if there were proper system call apis to accompany them. having to traverse a directory tree just to know how many processors are in the machine you are running on sysconf(_SC_NPROCESSORS_CONF); The_Franz posted:needing to parse a bunch of text from /proc/meminfo to find out the system memory stats is loving stupid and a pain in the rear end agreed on this. there was a syscall proposed for such behavior. in the gnome space we have libgtop but it's kind of dumb because its basically a bunch of autogenerated c code from perl
|
|
#
?
Jan 31, 2016 19:13
|
|
|
Suspicious Dish posted:it's kind of dumb because its basically a bunch of autogenerated c code from perl 
|
|
#
?
Jan 31, 2016 19:15
|
|
|
i think GNU is the missing element from this discussion about why linux is ubiquitous and bsd is dead (netcraft confirms) linux was more than software, it was the center of an ideology. it wasnt just about having a free unix, it was about FREEDOM. the revolution os. we joke about stallman/esr types but they were truly prolific characters. people actually believed in the mission. lol. what else could possibly explain people like suspicious dish and shadowhawk dedicating their lives to working on this poo poo for nothing/pittances? i mean most of that is over now, but bsd didnt have the critical mass of zealots and acolytes because the mission and philosophy isn't as 'visionary'
|
|
#
?
Jan 31, 2016 19:46
|
|
|
Suspicious Dish posted:sysconf(_SC_NPROCESSORS_CONF); that works for getting the basic number of cores available, but it doesn't help if you want more detailed info like how many physical cpu packages are installed, how many real cores vs hyperthreading, the clock speed, etc... osx is the only place where this is really easy since it's just a few sysctlbyname() calls. doing this on windows is an even bigger pain than linux since it involves counting bits in processor flags and digging into the registry. The_Franz fucked around with this message at 20:03 on Jan 31, 2016 |
|
#
?
Jan 31, 2016 19:49
|
|
|
Suspicious Dish posted:a bunch of autogenerated c code from perl eh, could be worse, you could maintain it by hand as long as the perl/w/e is a compile step and gets run and the output tested on a regular basis
|
|
#
?
Jan 31, 2016 20:51
|
|
|
Linux has a NUMA API and Windows has a similar beast for walking the entire tree of sockets, cores, processors, and threads. The number of physical installed, online, and active CPUs can be different.
|
|
#
?
Jan 31, 2016 21:00
|
|
|
I'm quite possibly the person *most* removed from the ideology. I use Windows on a regular basis, and I'm chatting with friends on Skype and have Photoshop open on my other monitor. I use proprietary software all day. My connection to this is literally "it's a hobby I picked up in high school, Red Hat liked my code and I lived 15 minutes away, and I saw it as an entry into a great software programming job" If Linux has a road in, it will be only when we stop caring about Linux. The most widely used deployment of user-facing Linux is Android, and they don't talk about the kernel at all. Use it because it's technically sane and has relatively good buy-in from vendors, but it's not magic and it's not a product all by itself.
|
|
#
?
Jan 31, 2016 21:02
|
|
|
i appreciate that linux exists but i'd never use it as my main operating system, it's too much of a pain in the rear end at times
|
|
#
?
Jan 31, 2016 21:07
|
|
|
Suspicious Dish posted:I'm quite possibly the person *most* removed from the ideology. 
|
|
#
?
Jan 31, 2016 21:11
|
|
|
does os x even support numa at all yet? i mean i guess there's no real reason for it to, because unlike linux (or even windows) it's basically not used for high-performance computing
|
|
#
?
Jan 31, 2016 21:41
|
|
|
there is not a single mac model with more than one cpu
|
|
#
?
Jan 31, 2016 21:46
|
|
|
|
| # ? Apr 20, 2024 02:58 |
|
|
pram posted:there is not a single mac model with more than one cpu The mac pro? 
|
|
#
?
Jan 31, 2016 21:51
|
|