Interview with Havoc Pennington of Red Hat 185
JigSaw writes "OSNews published an interview with Havoc Pennington, the head manager of Red Hat's Desktop department, also known for his freedesktop.org initiative and his very active/leading role in Gnome. Havoc discusses the internal changes on Red Hat, the future of the desktop version of Red Hat Linux, the XFree86 fork Xoutert, GTK+ and Gnome while he characteristically says regarding Linux eating UNIX's marketshare: '...nails are firmly in the UNIX coffin, and it's just a matter of time.'"
Linux vs Unix (Score:2, Interesting)
I think the momentum linux has over Unix is a matter of its GPL license, which makes it widely and freely available, and ensures a constant flux, NOT that it's technically superior. I think it's generally established that Solaris or FreeBSD for example, are technically superior to linux, however, both Sun's proprietary or BSD licenses are the detriments of those two Unices in the face of the Linux onslaught. (do you like how I used Unices as a plural to Unix? I do.)
Linux, ironically, now fits into the legendary "worse is better" more so than Unix does.
Re:Bad article style (Score:5, Interesting)
Because most online tech-site interviewers are *not* "grown up journalists," or even writers, and their operations are in fact run on shoe-string budgets which do not provide for in-person interviews. Consequently, when the interview is being conducted over the phone, through IM, or across several e-mail sessions, it's kinda tough to get a feel for what type of sofa upon which the interviewee is sitting.
Note, too, that most of the readers of tech-site interviews are not as discerning as you. They are looking for "news" or "answers" -- and quickly. No one browses OSNews in anticipation of savoring the linguistic bons mots of some proto-Hemingway.
How can a prediction be premature? (Score:3, Interesting)
Nor does the statement "it's only a matter of time" seem that outlandish. Already we've seen Linux being adopted by major Unix vendors. It's not unlikely that it will continue to make inroads in core areas and gradually drive their traditional offerings to more "niche" areas which ultimatly may be too small to warrent the expense of maintaining a seperate offering.
Re:UNIX is dying? (Score:2, Interesting)
More points:
Linux uses the "everything is files" philosophy, Win2K does not, Cygwin does not.
Linux has all "traditional" UNIX API:s.
Internally in the kernel, Linux has a lot of "traditional" UNIX solutions, tty/ptys, serial ports, filesystem mounts, networking, etc. is very UNIX.
Most utilities are UNIX-utilitie (cu, ls, mount, etc.).
Sysadmin stuff is basically UNIX (/etc/init, runlevels, inittab, and more).
Linux is therefore much more UNIX than Cygwin, BeOS or anything from Microsoft.
It just feels like UNIX in a way that other OSes does not. It therefore *is* UNIX in my book.
Not trademark UNIX, the Open Group defines that, but neither is Cygwin, BeOS or Win2K.
Re:Not that I have a bad attitude, but... (Score:3, Interesting)
Re:UNIX is dying? (Score:4, Interesting)
Just imagine for a moment what might happen if Sun released some version of the Solaris kernel under the GPL.
Solaris is far more scalable than Linux; Linux would begin bleeding enterprise market share, and would probably never recover. Solaris doesn't seem to scale down very well, so it is probable that Linux would retain its embedded systems market share.
The moment that Linux becomes a real threat to the software environment of an E15k, I have to believe that Sun will do whatever is necessary to protect this segment.
Of course, the sooner that Sun does this, the sooner that the question of enterprise Linux is put to rest. Sun is probably dragging its feet because they don't want to see Solaris running on the HP Superdome (especially since HP is killing their own UNIX and thereby depressing sales). They already have to contend with Solaris running on Fujitsu Primepower (which are arguably better machines than an E15k), but I am convinced that eventually, Sun will have to level the playing field by truly opening the Solaris source.
Re:Havoc Pennington has the right idea (Score:4, Interesting)
Or perhaps just someone with enough guts to put a foot down.
I finally got so sick and tired of doing tech support for family and friends that a couple of years ago I simply declared to them all, no more computer help for Microsoft products. I don't get paid for tech support, and Bill has too much money, so go bug him instead.
There was one person with a clue. My mother-in-law switched to Linux. I installed it for her and she finds it a pleasure to use compared to the constantly broken dozebox she sits in front of at work. It never has any significant problems, and when she does have the occasional question or concern it's a reasonable one that I can help her with.
Would she be able to compile a kernel? Shut up idiot, that's a stupid question. She doesn't have to. She turns it on and It Just Works (tm). And when the latest dozeworm comes wriggling across the wires, I don't have to drive 100 miles to her house to patch things up, because, like my own computers, hers is immune.
So yes, it's doable, and no, it doesn't mean you're a terrible friend or relative. Friends don't let friends use broken Microsoft products. And for that matter, true friends and family members don't try to press each other into involuntary servitude doing tech support.
Re:UNIX is dying? (Score:4, Interesting)
This is happening now, though not with Sun/Solaris... SGI's Altix handles up to 64 processors on a Linux kernel using the patches they release as opensource. As SGI hacks away at their bigmem and numa patches, they'll be able to handle more and more processors. The plan is to eventually graft enough IRIX technology to support just as many processors on Altix as they do with MIPS processors in Origin with IRIX.
Even if you aren't a fan of Itanium2, Linux, or NUMA, these patches are bringing some nifty high-end tech to the free software arena.
Re:UNIX is dying? (Score:2, Interesting)
Solaris also sucks hairy moose cock on Intel.
Not to mention that Linux performs better on the "low end" (read 2 to 4 CPU) configurations that dominate the computer market. Is "enterprise scalability" important? Yes, very very much so, but only in a tiny tiny fragment of the enterprise market.
Re:Havoc Pennington has the right idea (Score:2, Interesting)
Things like this are great for Linux penetration, 'cause when someone rings their ISP saying "My computer won't start up properly, it states that ntoskrnl is missing, and I don't have the CD or windows key", rather than saying "too bad call back when you have the CD", the ISP support staff can prod these "Mom & Dad" users in the direction of Linux.
Not to nitpick, but if someone called about ntoskrnl missing, (s)he would probably be redirected to their OEM or Microsoft to get that fixed. ISPs don't mess with installation problems.
Standardising (across distros) things like the location of the pppd configuration would allow (again, for the sake of this example) ISPs to provide quick training to staff on how to support Linux users ... Not possible when the ISPs position is that Linux can't be supported and staff who try are wasting company time.
Standardization would have been a huge help back when I worked with ISPs. However, at the ISPs I worked at, my previous co-workers had initiative.
At Mindspring, most of the help was located on a webserver, complete with links and images to simulate buttons and interfaces. They could "simulate" any version of Windows from 95 on ( since 3.11 wasn't supported ). This worked for Linux-based benefits: someone had created instructions on how to get PPP working with kppp and RH's ppp-config ( which I now forget the name of ) so a new user could connect to the dial-up servers. If you didn't have either, no worries, we could email ( if you had access via a friend ) or simply tell you the address for the PPP Howto.
At the other ISP I worked with, now shut down, Linux was supported by the NOC team. Since there was one present 24/7, any of the tech support folk could do their best ( most were Linux users ) at getting the customer online. Failing that, they'd call one of us, and we'd talk the customer through the process. Usually, by that point, it was less likely a PPP config error and more hardware related.