I was part of a Windows system install in 1993 for a mid-sized corporation. At that time, Windows 3.11 was wonderful. People were amazed by it. There were no writable CDs back then (floppies ruled), but it was being researched and people were excited about the huge storage potential. Things have changed a lot since that time, but some things have stayed the same (red black trees are still red black trees). MicroSoft is still a heck of a systems programming shop, but back then, they were God.
The Linux name change is a fitting tribute to Windows system programmers everywhere.
It is not a tribute. It is obviously a sarcastic joke cued by the coinciding version numbers. I am not sure how good Microsoft was at systems programming at the time, but I guarantee you Linus has very little regard for Microsoft's systems or any other type of programming acumen and is not about to start making tributes to Microsoft.
An entrepreneur can recognise a competitor without it being an attack of some form. Why does it have to be sarcastic? It's just a humorous acknowledgment of coinciding numbers - "Hey, '3.11' is a number that has significant history elsewhere in the computing world"
Certainly 3.11 is more evocative a number in that context than 10.4 or 6.1 or 3.0.
Obviously the version number is the reason, but tribute vs sarcasm, that's tough. It lacks any actual sarcasm, so I think Linus might have inadvertently made it a tribute, by mistake.
While I like linux, and I am writing this on an Ubuntu 13.10 install, there's really nothing very innovating about Linux, for anyone that has a unix history, it's just another variant, like the myriad of other ones [1].
You're setting quite a high bar for innovation if anything that "looks like unix" is not deemed to be innovative.
There's plenty of innovation in every modern "unixlike" OS. IllumOS, Linux, FreeBSD, NetBSD, to take some common examples, have each taken different paths in different functional areas, and rarely is one objectively worse than any other unless you have the constraint of a particular set of hardware to run the kernel on and/or software to run under the kernel.
Filesystems, for instance. ZFS and Btrfs (to a lesser extent, mainly because it's years from being feature complete) are innovative compared to earlier mainstream filesystems. Yet, a typical desktop user would not notice any innovation. Typical users are not concerned with snapshots, sending snapshot diffs between devices, or scrubbing a raid array for errors. ZFS has been criticized for being a resource hog, and both have been criticized for being slow. There are always trade offs. Getting a modern filesystem with COW and block checksums or hashes has performance costs, particularly on rotating disks. The only thing a desktop user will notice about a 5 year old computer running linux or BSD with a ZFS or btrfs filesystem is that it's slow. Slowness isn't innovative, right?
>there's really nothing very innovating about Linux
>Of course there's kernel innovation
:\
Perhaps ease up a bit on the hyperbole, you just took 2 polar opposite positions in the space of as many comments.
>ZFS - Linus had nothing to do with that
harshreality wasn't saying that Linus had anything to do with ZFS, he was noting it as an example of 'innovation [in something] that "looks like unix"' in order to counter your odd implication that that all unix variants are pretty much the same and devoid of innovation. To break down your argument, you say that:
1 - "Linux isn't innovative"
2 - "It's just another variant"
3 - "Like the myriad of other ones"
So Linux isn't innovative, and since all of the other ones (Solaris, BSDs, etc.) are alike, that they are neither innovative. You basically denied that there has been any innovation in Unix since it left Bell Labs. The ZFS example was a counter to that implicit claim.
>While I like linux, and I am writing this on an Ubuntu 13.10 install, there's really nothing very innovating about Linux, for anyone that has a unix history, it's just another variant, like the myriad of other ones [1].
I really wish I could use control-F on paper books.
I don't remember what book it was in, I think maybe the Art of Unix Programming or Just for Fun, but somebody remarked that the real innovation of Linux was the modern open source model of software development. And I think that's mostly true.
Your second paragraph is off topic and obviously wrong. I will not go into why it is wrong because it is off topic, but I cannot let patently false throw away statements stay unchallenged. Some poor soul may actually believe you.
Regarding your first paragraph, you cannot make a tribute inadvertently. The definition of the word tribute requires a certain intention. In layman's terms, if Linus did not intend something to be a tribute, then it is not a tribute.
Huge? Seriously? It's just a more efficient window manager. I don't really care much about the window manage, X has never been inefficient for me, my hardware is plenty fast enough. I think Mir is more likely a benefit for low end hardware like smartphones.
Window Manager used to have a very specific meaning, at least for me... WindowMaker, Sawfish, BlackBox, E16, E17, FVWM, TWM, these are all window managers. X11 is not a window manager. All window managers run under X11. These are the precepts.
Windows had a window manager too, and you could swap it out. There weren't nearly as many options.
Anyway, I for one am excited about Mir, because it means that my tf-101 or the replacement some years in the future may rejoin the convergence zone. They are making steps to live in the Android ecosystem, at least according to the article summaries I've come across that give any inkling of information about Mir.
If you google ASUS Transformer Ubuntu and follow some of the howtos you find, it's an absolute joke trying to get a windowed environment that resembles any netbook, laptop, or desktop you could buy last year or the year before that. "Now use Android VNC client to connect..." WTF?
Nobody else seems to be heading back to the convergence zone. Maybe wayland will get there at the same time. That's my rant.
Not that the definition of Window Manager is at all germane to the point, which I think was "I don't like efficiency".
Could have been either that, or "competition is bad, we already have the best way." You're right I think, it is probably characteristic of a manager to try to tell people what way to spend their time.
I don't think it lacks any sarcasm. Think about it. "For Workgroups". Wasn't that the new thing for Windows? Windows never was build with networking in mind. Heck, it wasn't even build with a multi user setup in mind.
All that came tacked on later and thus had inherent security issues because of backward compatibility.
So, calling Linux (which was always build around both, multi user and networking) "For Workgroups" in this 3.11 iteration is 100% sarcastic. Seeing as they (Linux) wouldn't even have to add that "For Workgroups" as it goes without saying that Linux is capable of it.
Linux was never intended to be truly innovative. It was intended to be a free/open unix. There was a strong demand for an open source unix that could be expanded upon, and there was none, until Linus released linux.
There was minix, which was a lot more innovative (microkernel), but Tanenbaum wanted to keep it strictly for educational purposes, simple and easy, and didn't want to turn it into a big system with all the bells and whistles that linux got.
Had FreeBSD been released earlier, that might have had the position that linux has now.
Right, like how there's nothing innovative in automobiles now; they're all still on four wheels and controlled with a steering wheel and pedals. I mean, come on!
Well...to be fair there really isn't. Perhaps in some new high tech Japanese supercars but the average car rolling off the assembly line in Detroit is about as innovative as a ham sandwich. Many cars still have incandescent light bulbs in them and the latest in-car tech is less innovative than 90% of the tech you find at a Toys-r-us store.
Although Windows for Workgroups was largely considered a joke (Windows for Warehouses) I also remember it being pretty wonderful. It brought to Windows all the SMB stuff that I enjoyed from OS2 as well as an actual built-in TCP/IP stack. I was happy to ditch WinSock.
Indeed. In the early 90s, it wasn't necessarily obvious that everybody needed a TCP/IP stack. "The Internet" was still essentially academic, and something the average person didn't really know anything about. Small PC LANs typically ran on some combination of IPX/SPX and NetBIOS.
3.11 laid the foundations for modular Windows networking, but included TCP/IP support was two years away in Windows 95, and even then it still wasn't installed by default(!). Microsoft took a while to catch on.
Attachmate merged with WRQ in 2005, and I believe they still sell Reflection branded products that do UNIXy things, like their Windows-based X server Reflection X.
Thank you for not making me feel like such an oldtimer here. As soon as I saw the name I thought "I bet I'm the only one who remembers it".
windows 3.11 aka "windows for warehouses" was actually pretty nifty and in a prior life I installed it at many small businesses who took full advantage of it's built in network mail client and many other very useful features.
It absolutely killed Lantastic which, at the time was pretty much pervasive in small businesses. Also at about the same time we replaced a lot of Arcnet and Token ring networks with UTP ethernet which I would have been amazed back then to hear would still be widely in use in 2013.
Wow. I think my first run-in with Windows 3.11 was in high-school -- and after having lived for years with an Amiga 2000 -- Windows was a complete joke of an OS.
I do remember having fun typing in single letters to files, and renaming them "somethin.com" -- I seem to recall eg: a "b" would predictably crash the computer instantly.
Not too long after that, as the pc hardware caught up with, and surpassed the Amiga, I got started with Slackware GNU/Linux. I find it hard to accept that anyone would seriously consider Windows 3.11 a good example of systems programming...
I'm not sure I would describe a team that releases an OS without memory protection, pre-emptive multitasking, or a 32-bit API several years after the introduction of the 386 a "heck of a systems programming shop".
The NT kernel, developed around that period, was good for its time though. Windows 3.x ... not so much.
I hate retrospective analysis of computing history which try to criticise things without understanding the context/reasoning, and instead look at it in a modern lens (e.g. people that call Windows 95 "bad").
You just contrasted Windows NT (1993) with Windows 3.0 (1990). Apart from the fact that NT is a MUCH newer OS released later (with 3.1/3.11 being upgrades onto an earlier OS), the system requirements of the two are quite starkly different.
For one example, Windows 3.0 runs on 384K of RAM. Windows NT 3.1 runs on 12 megabytes. Windows 3.0 also runs on 8 and 16 bit CPUs, while NT is 32 bit only.
If they had released Windows 3.0 as a 32 bit OS Microsoft literally might not exist as they do today, since Windows 3.0 was targeting cheaper IBM clones which were very behind the curve technologically. 32 bit would also have meant perhaps doubling the minimum memory requirement of the system.
Consider systems of the same era: A 1985 Amiga had pre-emptive multi-tasking on a system with a 16 bit data bus (though 32 bit internally), with 512K (technically 256K, but really it was not realistic to use a system like that very well). By 1990 there were plenty of networking solutions. There were also virtual memory extensions (but not memory protection).
It was bad for the era. It did well because the competing solutions only ran on expensive single-source hardware and/or were not seen as serious enough (even the big-box Amiga's were, aside from cost, negatively affected by being connected to the "game machines" on the low end).
This is not a modern lens. People back then also called it bad. I remember when Windows 95 was released, people mockingly called it MacOS 89, RiscOS 88, or AmigaOS 85.
I wonder who these "people" were. In 1990 I was using Apollo and HP workstations that made win3 look like a toy, but these were extremely expensive, high-spec devices. People who were accustomed to such systems may well have "called it bad". But for those with ordinary PCs at that time, windows 3 was perceived very differently.
MacOS and RiscOS at least were still going in 1995. Not to mention OS/2 was around. There really were better alternatives to both 3.11 and 95. A shell sitting on top of DOS was always quite hacky. They were big advances for people in the IBM compatible ecosystems who needed DOS compatibility, but not in a general sense.
Windows 3.0 also runs on 8 and 16 bit CPUs, while NT is 32 bit only.
There isn't a version of Windows for any 8-bit CPU. It did, however, run on an Intel 8088 which is a 16-bit CPU with an 8-bit external data bus. (Until Windows 3.1 came along which dropped real address mode support.)
Unfair. Windows 3.x had to be able to run on an 8086 (real mode), because that was what most people had. It could also make use of a 286 (standard mode) or 386 (enhanced mode) if available.
Given the hardware constraints I think it was an impressive piece of work for the time.
My first experience with Windows for Workgroups was in 1994 when we were forced to use it at work. By that time I'd already been using Linux for a few years (since before 1.0).
In my opinion, WFW didn't really hold a candle to X11 and Linux, and the networking options were frankly laughable. Even with Linux's crummy NFS implementation at the time, it was far superior to using Netbeui/IPX and connecting to a Novell Netware server.
"At that time, Windows 3.11 was wonderful. People were amazed by it."
These must have been people with little computer experience.
In '93 I also ran 3.11, at home. At work I had access to Big Iron, VMS, DG UX, and a hand full of other OSes. There wasn't a really decent Microsoft OS until Windows 2000. Microsoft was only god on the desktop because it allow the average user tools like Office. The engineers and professionals used Sun workstations and Macs, production used VMS, which was decades ahead of any of the PC OSes at the time.
Personally, I don't think it was all that wonderful, particularly when WfW 3.11 required a 386 anyway. This is part of why I hate the MS OS/2 2.0 fiasco quite a lot.
As a Mac user at the time, the love for Windows 3.11 left me scratching my head something fierce. By comparison to System 7.1+, Windows 3.11 was awwwwful.
Of course, around that time you could also run an Irix desktop, complete with Photoshop and SoftPC and even (later on) IE, but that was sadly beyond the reach of most consumers.
Not sure why you would say this. My first job was running phone lines through the lowered ceiling to hook up localtalk for the Macs. It worked wonderfully. Even back then Apple was very auto-configure-y which made finding printers and file servers very easy. Just go into Chooser and there they all were.
IP over localtalk was not difficult either. After a couple years it moved to ethernet (coax, yay!) and that worked fine too. Networking Macs has always been extremely easy.
I agree. The Mac was great at networking in 1993. Sure, MacTCP and MacPPP may have had strange configuration UI but localtalk and ethertalk Just Worked, and getting MacTCP and MacPPP working wasn't too hard.
My memories of AppleTalk wasn't so fond - though I can't for the life of me remember why as it seems so long ago now and I've filled my brain full of a few decades more technojunk since. But I do remember being joyful when it came time to port the various Mac LANs over to TCP/IP.
Both Mac and Windows may lagged behind the various unixes at the time in terms of WAN technologies (such as TCP/IP), but the Mac had fantastic LAN capabilities in the form of AppleTalk. Auto-discoverability and auto-setup was a dream. Just give your Mac a name, enable sharing, and you were off and running! Anyone on the network could find your stuff and share with you. All done with human-readable names and very simple GUI tools.
There was even good cross-subnet capabilities in the form of AppleTalk "zones", although you had to have routers which could speak AppleTalk in order for it to work. But when it did work, man was it nice! Far ahead of Windows at the time, and still superior in many ways when it came to ease of use vs. the zeroconf stuff of today.
IIRC, almost everything Windows 3.11 had out of the box was already present on Macs for ages. Shared printers, p2p file servers, instant messaging (that was an add-on) and more. And better - it all just worked.
I remember that, at the time, the e-mail client/server that came with 3.11 was nice and I don't remember anything like that on Macs, but, with that exception, System 7 was orders of magnitude better than 3.11.
My memory might be fuzzy, but I remember support in those days being very limited for 7.1. Win 3.11 would run on stuff made in the 80s (any 386 or 486 with enough RAM) - OS7 required a fairly new Mac.
3.1 and 3.11 were far different than the 0.01 would suggest.
* "Standard mode" was dropped, so 3.11 required a 386 or above
* Significant networking improvements (hence the "for workgroups" in the name) including IIRC built-in TCP/IP support (prior to that most people used Trumpet Windsock)
* I/O subsystem improvements too
I've heard it said that WfW3.11 was essentially a huge gamma-test for the network and I/O subsystems intended for what became Windows 95.
There was a non WfW 3.11 though that I think didn't include VSHARE.386; by the time 3.11 was around I think I had my 386/40
Also, WfW did not initially ship with built in winsock implementation, WfW did ship all the networking apparatuses other than that, just needing a drop-in winsock.
"Perkeleen" is used nearly synonymously with "fucking" though (as in, "devil" isn't really considered swearing in the English language). I read it as "fucking cunthead". Which is lovely indeed!
Disclaimer: my knowledge of the Finnish language is mostly limited to profanity.
I'd rate 'perkeleen' to be slightly 'softer' than 'fucking'. 'Vittu' (or 'vitun') is the best translation for 'fucking'. Not sure if there's a better word for it in English, but maybe something like 'damn' or 'bloody'.
Charming gestures like this is part of the reason I like the open-source community.
You don't see nonsense like this very often in commercial products (except maybe the Goats transported system-counter in Android, whose open-source status is constantly under debate).
As a Windows user, I'm always surprised to see how conservatively Linux uses its swap area. On Windows, even with enough free memory, it still pages to disk and gives me a sluggish experience.
Does anyone have any idea why Windows is so aggressive about paging disks out compared to Linux?
I can't explain the sluggish experience, but a plausible explanation is that Windows sees a bunch of unused pages, and also a bunch of uncached disk blocks that are used often, and decides that it would be preferable to put those disk blocks in memory in exchange for putting the "unused" memory in swap.
zram in its various incarnations is awesome and has been available in Linux for a few years at least now. It's interesting that Apple is copying it.
Linux gives lots of tuners to adjust swappiness. As far as I know, Microsoft assumes that one (conservative) setting fits all, in typical Microsoft fashion.
I'm still on Ubuntu 12.04 LTS which has kernel 3.2.0-49, I thought 'old' was 2.6 or really old 2.4. What the heck is going on now, the kernel is already up to 3.11? Did I miss anything good in between? By my perceptions I would have imagined that newest kernel would be around 3.6 or so but I was wrong: https://www.kernel.org/finger_banner
2.4 is really old, originally released in 2001. 2.6 has also been around for a long time, and the release cycle changed during the 2.6 timeframe, so the old release number system no longer makes sense.
Rather than having a major stable release like 2.4, and then a long period of development in 2.5, followed by a stable 2.6 kernel later on, it switched to much shorter merge windows and release candidate cycles that last about 2-3 months. So you had 2.6.x kernels which were new stable branches, and would get patches such as 2.6.x.y.
After a while, this got a little silly. When numbers were getting up to things like 2.6.32.60, it gets hard to keep track of the numbers. Since the first two numbers were just sitting there unused, at some point Linus decided it was silly, and they would cycle to 3, and the second number would be for each stable release.
So yeah, 2.2-2.4 was about 2 years between new stable versions. 2.4-2.6 was about 3 years between new stable versions. But then there have been new stable versions about once every 2-3 months, and new long term supported versions picked from those about once a year (for distributions which prefer to do long-term support on a single kernel version, rather than updating to the latest stable).
> After a while, this got a little silly. When numbers were getting up to things like 2.6.32.60, it gets hard to keep track of the numbers. Since the first two numbers were just sitting there unused, at some point Linus decided it was silly, and they would cycle to 3, and the second number would be for each stable release.
And now we have only one number sitting there unused! Maybe they'll finish the job next time.
The kernel numbering system changed when Linus moved over to v3. Basically the switch from 2.4.x to 2.6.x was the last major change to the Linux kernel. Since then, all the 2.6.x releases have been pretty standard incremental updates. So a few months ago Linus decided it was time to switch to v3 - just to round the numbers off (it wasn't a big nor noteworthy release - it was just an excuse to round the numbering off). So now the the second digit is what used to be the 3rd digit in the 2.6.x releases.
It usually takes distro developers a while to vet kernels and ensure stability with their userspace stuff. 3.7 and 3.8 had a few regressions, and were skipped by a few distros. 3.9 only made it into Debian Stable last month (as a backport from testing).
On Ubuntu, you can always get some newer kernel versions through apt-get, but the default will always be what they consider to be stable (especially for an LTS)
If you want to move to the newer 3.5 kernel line, you can do so. This is what the 12.04.2 LTS is distributed with. Here is the apt-get for that:
sudo apt-get install linux-image-generic-lts-quantal
You will also want to update your x.org install:
sudo apt-get install xserver-xorg-lts-quantal
Note that some older video cards (nvidia I think) have had their support dropped, so this operation may or may not be safe.
Thanks for the suggestion, I found a top level Quantal package and that pulled in the headers too in case anything needs to compile something.
Also, I looked at the X stuff too but it wanted to remove ubuntu-desktop and that seemed pretty drastic so I'm just using the newer 3.5 kernel for now. On this new PC I picked up support for my hot new 1 Gig ethernet, everything else worked fine but now I have an excuse to upgrade to a 1 Gig home router. :) Well once the cable can go fast enough that is. Assuming Google Fiber gets here, they promised Austin but Round Rock might not get it.
Just an interesting observation that both of these "3.11" are actually referring to different relative versions.
Windows was not using the now-common semantic versioning. Windows 3.11 is actually a patch to 3.1, which in semantic versioning would actually be 3.1.1.
Linux 3.11 however, is a minor upgrade to Linux 3.10.
Welcome to Hacker News! I don't mean for this to come off the wrong way, but please try to put more effort into your comments than Imgur links. This is a place for discussion, not one-liners and images. It may be that way on Reddit, but Hacker News has a focus on higher effort content.
The Linux name change is a fitting tribute to Windows system programmers everywhere.