Re: Linux or GNU/Linux
| From: | Alan Cox <alan-AT-lxorguk.ukuu.org.uk> | |
| To: | Community support for Fedora users <users-AT-lists.fedoraproject.org> | |
| Subject: | Re: Linux or GNU/Linux | |
| Date: | Fri, 3 Aug 2012 14:36:06 +0100 | |
| Message-ID: | <20120803143606.05979a4b@pyramind.ukuu.org.uk> | |
| Archive‑link: | Article |
> However, the kernel was far less successful. Despite the claims > of "superior architecture", Hurd has never caught on; almost every > implementation you'll run into (I'd say 'every', but someone would come > up with John Drutin's Distro as a counter-example) uses the Linux kernel > and GNU libraries and utilities. Read the Wikipedia article on "GNU Hurd". To be fair to the FSF there is often a perception that the FSF screwed up here. In a sense yes they made some wrong decisions, and then very stupidly were insulting and hostile to the Linux authors but there is a background that makes more sense. When the GNU OS concept started the idea that everyone would have a Unix capable system on their desk was pretty hard to imagine. The choice of a Mach based microkernel was both in keeping with a lot of the research of the time and also had a social element. The vision was a machine where any user could for example implement their own personal file system without interfering with other users. Viewed in the modern PC world that sounds loopy but on a shared multi-user computer it was an important aspect of software freedom. Sticking to Mach and being hostile to Linux wasn't very smart and a lot of developers have not forgiven the FSF for that, which is one reason they find the "GNU/Linux" label deeply insulting. The other screw up was that they turned down the use of UZI, which would have given them a working if basic v7 Unix equivalent OS years before Linux was released. Had they done that Linux would never have happened and probably the great Windows battle would have been much more fascinating. > Richard Stallman wants people to acknowledge the contribution GNU has made > to the success of the Linux distributions by refering to them collectively > as "GNU/Linux". People tend to go for the shortest term that conveys a > meaning, so the vast majority of people simply call it "Linux", and those > who know understand that the GNU utilities, commands, and libraries are > included. In some cases - by code volume the FSF is not the biggest contributor, even when you include (as they like to) all the third party code they took and labelled GNU. In the case of the largest and most common Linux distribution today there is almost no GNU code in it: Android. Alan -- users mailing list users@lists.fedoraproject.org To unsubscribe or change subscription options: https://admin.fedoraproject.org/mailman/listinfo/users Guidelines: http://fedoraproject.org/wiki/Mailing_list_guidelines Have a question? Ask away: http://ask.fedoraproject.org
Posted Aug 9, 2012 1:33 UTC (Thu)
by mathstuf (subscriber, #69389)
[Link] (2 responses)
Posted Aug 9, 2012 2:19 UTC (Thu)
by malcolmt (guest, #65441)
[Link]
Posted Aug 9, 2012 4:24 UTC (Thu)
by mlinksva (guest, #38268)
[Link]
Also quoted in https://lkml.org/lkml/1996/4/22/21 "BTW when I was first on the net there were all sorts of 'hurd is too ambitious we should base it on UZI' type discussions." Too bad those aren't indexed, AFAICT.
I'd enjoy reading a history of "[FSF] turned down the use of UZI".
Posted Aug 9, 2012 23:42 UTC (Thu)
by dvdeug (guest, #10998)
[Link] (6 responses)
Posted Aug 10, 2012 16:30 UTC (Fri)
by daniel (guest, #3181)
[Link]
Posted Aug 16, 2012 2:30 UTC (Thu)
by scientes (guest, #83068)
[Link] (4 responses)
Posted Aug 21, 2012 14:31 UTC (Tue)
by viro (subscriber, #7872)
[Link] (3 responses)
Posted Aug 21, 2012 22:28 UTC (Tue)
by nix (subscriber, #2304)
[Link] (2 responses)
More generally, arbitrary limits are bad (unless essential for stability or as sanity limits: the kernel's various symlink-chase restrictions count as both). ld.so is in userspace and should therefore be relatively free of arbitrary restrictions (no harsh stacksize limits nor need to avoid DoS attacks in this domain, as a binary that has so many shared libraries that it takes ages to load affects nothing but itself). So I'd assume a limit on loadable shared libraries imposed only by the VMA limit in the kernel. Anything less than that is just shoddy engineering.
Posted Aug 21, 2012 23:58 UTC (Tue)
by viro (subscriber, #7872)
[Link] (1 responses)
Frankly, library dependency trees are disgrace; more often than not it's a shoddy engineering in its own right... FWIW, the fattest I see here is qemu, with its shitloads of alternative sound, etc. libraries linked in (as well as ridiculous crap like libdbus, really funny on a box that isn't dbus-infested - Not On My Damn Network, TYVM)... And it still amounts only to 63 shared libs.
In principle I agree that this linker limit is wrong, but if we were to apportion the blame... This is far stronger indication of lossage in openorifice and its forks.
In any case, I strongly suspect that the real reason why they abstained from touching the GNU ld had very little to do with the license and a lot - with the inherent beauty of ld codebase and its incredible efficiency in the hard tasks of wasting cycles and RAM. IOW, the same reasons that had resulted in creation of gold(1) several years later. It got started 3 years too late to be of any use for android folks and hadn't grown ARM support for another year. Add a bit for stabilization and at that point switching the linker would probably have been too painful to justify.
Hell knows - ask Google android folks; they show up here from time to time. Or ask other Google folks, for that matter - IIRC, Ian Lance Taylor is also there...
Posted Aug 22, 2012 13:13 UTC (Wed)
by nix (subscriber, #2304)
[Link]
As for library dependencies, my fullish vim reports libgtk-x11-2.0.so.0 libgdk-x11-2.0.so.0 libgdk_pixbuf-2.0.so.0 libpango-1.0.so.0 libgobject-2.0.so.0 libglib-2.0.so.0 libgnomeui-2.so.0 libbonoboui-2.so.0 libgnome-2.so.0 libSM.so.6 libICE.so.6 libXt.so.6 libX11.so.6 libm.so.6 libncurses.so.6 libacl.so.1 libdl.so.2 libperl.so libpthread.so.0 libc.so.6 as direct DT_NEEDEDs. The rest are pulled in as indirect dependencies.
(Most distros will have full vim as an extra package, because gvim pulls in half the known universe via its GNOME dependencies -- this probably explains the discrepancy. I suspect my Emacs would exceed this count, except I'm still using an Xawed Emacs because of an ancient unfixed bug in Gtk whereby when you disconnect from an X server, you'd better exit() or Gtk will crash you shortly afterwards, because they can't be bothered to track which in-memory resources come from which X server. Since Emacs supports --daemon and can connect to many distinct X servers over the course of its lifespan via the emacsclient, this bug is somewhat lethal to it.)
Posted Aug 16, 2012 14:36 UTC (Thu)
by landley (guest, #6789)
[Link] (1 responses)
Hard for who to imagine? In 1983 we 8-bit kids already had Moore's Law in our bones: the difference between the Timex Sinclair, Vic-20, and Commodore 64 drove that home pretty hard. Our model of the work environment's we'd grow up to use was the bridge of the original Enterprise. When the C128 came out the joke was "C64, C128... C3P0".
Failure to expect people to have hardware _capable_ of stuff would have meant he wasn't asking people young enough.
Of course we 8-bit kids were too young to care about Unix, but before the PC shipped Paul Allen (the Steve Wozniak of Microsoft) licensed Unix from AT&T on the _rumor_ that IBM was doing a PC, and hired a 2 guys in a garage operation called "SCO" as consultants to port it to the two potential CPU candidates (8086 and 68k) IBM was rumored to be considering, and called it "Xenix" to say he'd happily port it anywhere else IBM needed it if he'd guessed wrong. When Allen finally signed the NDA and found out they were doing 16-bit S-100 variant (ISA was electrically just S100 with unused wires removed) that maxed out at 64k ram in the first iteration, he bought a CP/M clone (from his Seattle friend who'd previously done an 8086 card for the Apple II from Microsoft to run its BASIC on) and made DOS out of it, but then in DOS 2.0 Allen shoehorned as many Unix features in as he could to form a transition path to Xenix. His plan (explained in the DOS 2 manual) was to ship Xenix as DOS 4 (if he hadn't come down with Hodgkins Lymphoma in 1983 and quit Microsoft after overhearing Gates and Ballmer talking about how to get his stock back if he died, this might have worked).
Stallman really didn't do anything visionary, he was forced to react to the death of his beloved PDP-10 system after Jupiter got cancelled.
Rob
Posted Aug 17, 2012 5:13 UTC (Fri)
by drag (guest, #31333)
[Link]
On a side note:
Of course everybody should understand that Microsoft got it's start as a Unix vendor. If I have my history right Microsoft's first commercial operating system was Unix, not DOS. One major market was a popular line of point of sale devices that was used by various nation-wide chains in the USA. Companies like Blockbuster or Pizzahut used it as the OS for tracking orders and inventory. In the 386 days SCO was hired by Microsoft to develop Xenix on their behalf. Unix was also the primary development platform used by Microsoft up until Windows for Workgroups; around the early to mid 1990's. Microsoft lost interest in Unix with their work with IBM on the OS/2 effort, which was then superseded by NT.
Also Microsoft's Unix was probably was the most widespread Unix version of it's time. Microsoft showed significant foresight by abandoning the Unix morass and striking it out on their own. Also they managed to avoid IBM sucking them back in.
Another side note... The SCO of today is not the same SCO that hacked on Xenix and developed OpenServer. That SCO was effectively destroyed by their relationship with IBM. They threw themselves on the rocks with their attempts to collaborate with IBM over a unified Unix platform. This was a commercial failure and IBM happily used Linux as a much more viable alternative. SCO lacked the resources to recover from this fiasco and their primary platform languished as a result. As a result People migrated away from it in ever larger numbers... SCO sold off it's Unix side in 2001.
The SCO of 'sue Linux' lawsuit fame is really 'The SCO Group', which was the company previously known as Caldera. They were one of the many commercial Linux vendors that tried the 'proprietary value added features' model.. which turned out to be a fatally flawed business model.
The remaining bits of the 'real SCO' really went off to try to be a web platform company... which was bought by Sun, which is now part of Oracle... which is now using Unix and Linux as a vehicle to fight off their current greatest nemesis: Microsoft.
My conclusions on all of this is:
All in all the history and rise and fall of Unix is a miserable tragedy. There are quite a lot of other things that happened that isn't mentioned above.. like Microsoft's using legacy contracts with SCO over Xenix code to try to damage that company's ability to compete with NT, or how the now infamous BSD copyright lawsuits went down.
If it wasn't for the industry's stupid obsession with IP (and the resulting willingness to throw themselves on that sword) we would all be probably running variations of 'real' Unix systems. Likely hybrid open source/closed source systems (although most people probably wouldn't think in terms like that) with a handful of different academic and industry working groups with full access to the code sponsored by vendors to continue the develop different aspects of the platform.
I guess the state of the art for *nix systems has been held back by about 20 years or more because of all of the legal mumbo jumbo and people pursuing the 'proprietary features' and 'proprietary add on' approaches to try to garner market share and lock in customers.
Re: Linux or GNU/Linux
UZI was Doug Braun's Unix for the (Zilog) Z80 chip.
There's an explanation here and code download available, too.
Re: Linux or GNU/Linux
Re: Linux or GNU/Linux
Re: Linux or GNU/Linux
Re: Linux or GNU/Linux
Re: Linux or GNU/Linux
Re: Linux or GNU/Linux
Re: Linux or GNU/Linux
Re: Linux or GNU/Linux
linux-gate ld-linux libgcc_s libc libm libdl libselinux libtinfo libacl libgpm libperl libpthread libpython2 libruby libattr libresolv libnsl libcrypt libutil librt libfreebl3
What the hell is yours linked against? nvi is libc/libdl/libncurses/libtinfo
plus the usual minimum (linux-gate/ld-linux)...
Re: Linux or GNU/Linux
Re: Linux or GNU/Linux
> capable system on their desk was pretty hard to imagine.
Re: Linux or GNU/Linux