|
Cocoa Crispies posted:lol glwt Literally thought this was a window manager or something like olwm
|
#
?
Nov 9, 2019 01:07
|
|
|
|
| # ? May 27, 2024 02:39 |
|
|
BobHoward posted:which brings us back to my original assertion: to most people hacking on linux circa 1993, sgi hardware and software was a legend they’d never see. it’s as if the most base Mac today was the new tower Mac Pro: in that hypothetical world, even if you as a linux hacker wanted to check out macOS and mine it for ideas, you’d never buy a Mac with your own money unless you were real rich or could make it pay for itself somehow feedmegin posted:Tbf you wouldn't OWN an SGI box back then but if you were a student - and lots of us were - you might well have access to one via your uni's compsci labs etc. I had serious envy of some of the Solaris workstations I got to see a little later in the 90s compared to my 486. 17" monitors with 24bpp! right so broke rear end tech nerds in the 1990s owned used gear, nobody was buying a brand new sgi indy to fiddle-fart around on at home i hope
|
|
#
?
Nov 9, 2019 02:21
|
|
|
Soricidus posted:please define “unix workstation” here. I assume you are somehow excluding things like the next and hp-ux workstations that were based on, uh, the 68040? yes 68k-based unix workstations were popular in the 1980s next was an attempt at a low-cost workstation, with a weird kernel to save money, and cheap chips, to save money. they went bankrupt trying to sell hardware no one wanted, so i'm gonna let that one alone. sun's last 68k system was released in 1989.... it was about 10% as fast as their entry level sparc system, but cost about 60% as much. you can see how that was a rough value proposition hp sold 68k workstations longer than sun, but not much longer. after they brought out their first "cheap" risc systems in 1991 I think it was over for the hp300 stuff.
|
|
#
?
Nov 9, 2019 02:29
|
|
|
BobHoward posted:^^^ i was thinking of mentioning that i was an undergrad student and we did have a sgi lab but it was mega elitist, reserved for grad students and maybe 1 or 2 specific undergrad courses. the rest of us couldn’t even get in (locked door) Legit wish your postan rig was kept in a lab to which you had no access (due to a locked door).
|
|
#
?
Nov 9, 2019 03:32
|
|
|
Notorious b.s.d. posted:hp sold 68k workstations longer than sun, but not much longer. after they brought out their first "cheap" risc systems in 1991 I think it was over for the hp300 stuff. HP mainly sold 68K stuff for legacy reasons in the 1990s: the 9000-400 series workstations were to help transition Apollo users to HP-UX and the final 9000-300 modular systems were mainly for extending existing deployments the 9000-300 series was made up mainly of boards and modular cabinets that you could configure in a bewildering variety of ways; you’d get some higher end ones, make HP-UX (or BASIC/Pascal) workstations out of them, then use those to develop process control software to run on lower end boards from either SRM or ROM to run all the systems in your factory
|
|
#
?
Nov 9, 2019 03:47
|
|
|
eschaton posted:HP mainly sold 68K stuff for legacy reasons in the 1990s: the 9000-400 series workstations were to help transition Apollo users to HP-UX and the final 9000-300 modular systems were mainly for extending existing deployments this was all before my time oldest hpux poo poo i ever touched was series 700, the line that mopped all this up and killed the two-streams hp-ux thing i know even those early cost-reduced hppa systems were a heck of a lot faster than a 486 or a 68040 though. and they ran nextstep better 
|
|
#
?
Nov 9, 2019 05:21
|
|
|
Notorious b.s.d. posted:right so broke rear end tech nerds in the 1990s owned used gear, nobody was buying a brand new sgi indy to fiddle-fart around on at home i hope Nobody was buying an old secondhand workstation at all as their home machine, why would you when you could get a cheap PC running this 'Linux' thing?
|
|
#
?
Nov 9, 2019 15:29
|
|
|
sure people were they were the ones who got the ball rolling porting Linux and BSD to their platforms because getting support from the vendors went from possible (“of course you can use the latest OS on your workstation at home, here you go”) to impossible (“you’ll need an annual support contract and to pay a fee for every upgrade”) in the mid-1990s one of the reasons Symbolics hardware is so rare is because they always treated people who wanted to take old hardware home that way; they preferred it be scrapped or returned to them and required a contractual agreement from a site not to operate old hardware—including people taking it home—if they reduced their support contracts
|
|
#
?
Nov 9, 2019 22:38
|
|
|
feedmegin posted:Nobody was buying an old secondhand workstation at all as their home machine, why would you when you could get a cheap PC running this 'Linux' thing? i am speaking from experience. i had a secondhand workstation as my home machine and i knew lots of others who did the same. secondhand workstations were often cheaper and faster than a pc running freebsd or linux or solaris x86. not to mention linux really sucked back in the day. smp didn't work and stability was hit or miss. remember linux 2.4, "the kernel of pain" ? yeeeeah.
|
|
#
?
Nov 10, 2019 00:27
|
|
|
my first linux system was a laptop, not even kidding. laptops were better supported under linux than ostensibly less-lovely unix systems, so i had good reason to go out of my way to figure out this "linux" thing and yes, sound worked on day one
|
|
#
?
Nov 10, 2019 00:30
|
|
|
did it still work on day two
|
|
#
?
Nov 10, 2019 05:41
|
|
|
Notorious b.s.d. posted:not to mention linux really sucked back in the day. smp didn't work and stability was hit or miss. "back in the day" yeah, sure.
|
|
#
?
Nov 10, 2019 17:40
|
|
|
i think the first linux i used was ubuntu 7.04 so no, i don't remember the kernel of pain.
|
|
#
?
Nov 10, 2019 18:39
|
|
|
i started using linux during 2.4 (mandrake) and it seemed fine even if a bunch of things were less good than they are now. that was in 2002 or something and nobody had more than one processor so it didn't matter
|
|
#
?
Nov 10, 2019 19:13
|
|
|
hell, I remember kernel 2.2.
|
|
#
?
Nov 10, 2019 19:13
|
|
|
His problem was that he was trying to use a laptop. Commodity PCs had full driver support for most hardware that wasn't a brand new release in 2.0 and 2.2.
|
|
#
?
Nov 10, 2019 19:25
|
|
|
i remember the gigantic jump in performance going from 2.0 to 2.2. gave new legs to my old second-hand-me-down p75 on which i played quake2 with the software renderer, i finally had stable 30+fps, up from 25 average
|
|
#
?
Nov 10, 2019 19:35
|
|
|
esd > arts amirite
|
|
#
?
Nov 10, 2019 19:36
|
|
|
James Baud posted:His problem was that he was trying to use a laptop. Commodity PCs had full driver support for most hardware that wasn't a brand new release in 2.0 and 2.2. no the laptop was fully supported under linux 2.0, it was pretty great that's why i was installing linux, remember? the solaris x86 laptop situation wasn't great, to say the least
|
|
#
?
Nov 10, 2019 20:10
|
|
|
rt4 posted:i started using linux during 2.4 (mandrake) and it seemed fine even if a bunch of things were less good than they are now. that was in 2002 or something and nobody had more than one processor so it didn't matter it mattered a lot, hence it being a huge development priority and the root of many issues
|
|
#
?
Nov 10, 2019 20:11
|
|
|
oh sure, for people with servers
|
|
#
?
Nov 10, 2019 20:40
|
|
|
rt4 posted:oh sure, for people with servers 
|
|
#
?
Nov 10, 2019 21:12
|
|
|
i remember xfs on kernel 2.4 it was guaranteed to destroy every single file that was open at the time of an unclean system shutdown good poo poo
|
|
#
?
Nov 10, 2019 21:22
|
|
|
Ipchains? Pepperidge farms remembers.
|
|
#
?
Nov 10, 2019 21:25
|
|
|
lmao if you don't fondly remember kernel 2.4 as being an incredible improvement over its predecessors
|
|
#
?
Nov 10, 2019 21:37
|
|
|
Sapozhnik posted:i remember xfs on kernel 2.4 lol how even
|
|
#
?
Nov 10, 2019 21:38
|
|
|
I also remember patching the Nvidia driver for kernel 2.5
|
|
#
?
Nov 10, 2019 21:41
|
|
|
psiox posted:lmao if you don't fondly remember kernel 2.4 as being an incredible improvement over its predecessors
|
|
#
?
Nov 10, 2019 21:49
|
|
|
mystes posted:If I remember correctly, the old versioning system completely sucked because 2.2 had gotten ridiculously out of date before 2.4 was released and so you had to use the unstable 2.3 kernel for tons of stuff. yeah :/ all rolling on the -mm tree i just discovered that software suspend now Just Works on the latest ubuntu on my surface pro 3 (which caused me to learn about how ACPI sleep states are barely a real thing anymore), so i am having a Happy With Linux kinda day
|
|
#
?
Nov 10, 2019 21:55
|
|
|
mystes posted:If I remember correctly, the old versioning system completely sucked because 2.2 had gotten ridiculously out of date before 2.4 was released and so you had to use the unstable 2.3 kernel for tons of stuff. same with 2.4/2.5, glad they stopped doing that with 2.6
|
|
#
?
Nov 10, 2019 22:07
|
|
|
Sapozhnik posted:i remember xfs on kernel 2.4 technically this never changed on the xfs side -- they never really re-architected the code to allow this (ab)use case. what happened was that the linux kernel got reliable write barrier support, and then the xfs devs were more than happy to plug into that, because that solution is sane, instead of trying to handwave your way around layers underneath you write barriers in linux are carried through all the way from the filesystem to actual block device drivers, so when your filesystem asks for a write barrier there is a reasonable chance that the right things will happen all the way down
|
|
#
?
Nov 10, 2019 22:40
|
|
|
Progressive JPEG posted:same with 2.4/2.5, glad they stopped doing that with 2.6 the 2.4=>2.5 transition was incredibly confusing because new drivers went into 2.4 all the time but the actual stable smp kernel that worked reliably was 2.5.99, which you had to compile yourself, because linux 2.5.99 / 2.6 was a tremendous leap forward and, along with threading/polling stuff, made linux actually competitive with "real" unix
|
|
#
?
Nov 10, 2019 22:42
|
|
|
remember the days when you actually had to compile linux kernels lol crazy
|
|
#
?
Nov 10, 2019 22:49
|
|
|
2.6 not incidentally also the last time i particularly cared about a linux release. things got sorted out enough that it wasn't that enthusiastic a thing anymore
|
|
#
?
Nov 10, 2019 23:09
|
|
|
Notorious b.s.d. posted:not to mention linux really sucked back in the day. smp didn't work and stability was hit or miss. who needs smp? threads are for people who can't program state machines
|
|
#
?
Nov 11, 2019 00:09
|
|
|
Sapozhnik posted:remember the days when you actually had to compile linux kernels lol I still do. Embedded Linux and all that.
|
|
#
?
Nov 11, 2019 01:46
|
|
|
it made me feel really old when I realized that many of my younger colleagues - who are professional sysadmins and/or SREs have never compiled a kernel and may never need to
|
|
#
?
Nov 11, 2019 03:18
|
|
|
its not hard. what are they missing out on. even has an ncurses menu which is scrub tier af imo
|
|
#
?
Nov 11, 2019 03:19
|
|
|
if youre a scrub, compiling a kernel and watching make spam is a great way to feel leet using outdated vocabulary to talk about an outdated practice is intentional
|
|
#
?
Nov 11, 2019 04:38
|
|
|
|
| # ? May 27, 2024 02:39 |
|
|
Achmed Jones posted:it made me feel really old when I realized that many of my younger colleagues - who are professional sysadmins and/or SREs have never compiled a kernel and may never need to sounds good to me op
|
|
#
?
Nov 11, 2019 04:44
|
|