For the moderately near future, I don't see a linux distribution locking out a person from supporting it themselves like MS does (though getting support from your linux/hw vendor may be complicated by his actions). However, I do think the installers plan and execution was not good for a business install. Actually, it's not that good for home installs either.
One thing:
its patsies, HP In this case, HP is not just being MS' patsy, they are serving their own interest. MS didn't request them cripple their customization, they crippled it themselves to encourage any upgrades to be HP supplied, not third party.
The key sentence where everything went to crap:
Out came the anaemic 40GB drive from one HP, and in when the Promise controller and two WD 200GB SATA drives. In the first part, he drank the vendor kool-aid and got their customized XP install. That's a valid choice in and of itself. Then he grabbed a third-party controller card and expected the stuff HP provided to play nice with it. HP didn't want him to do that, they wanted him to buy a presumably much more expensive HP branded controller to do the same thing. HP's install CD not accomodating that is hardly a surprise, and hardly a MS decision. The different keys for retail and OEM reflect the different pricing tiers. .
If they are a particularly small business, not going with one vendor is a valid choice, but you best put it together via all-third-party parts and get a generic OEM windows disk. If you can get a no-windows discount on the HP system, and use that discount for a different license, you can go with a non-restricted install media set. You do, however, in this way accept a higher degree of risk (problem determination falls squarely on your shoulders, and your vendors may disagree with your conclusion and blame other parts..). If you run on thin margins and time is not uber-critical for systems, this may be the appropriate path
If you drink the vendor kool-aid and get their hardware and software, you've drunk the kool-aid and as a consequence, you ought buy from HP your upgrades. You can't expect something put together by them to work for hardware configurations they would explicitly not support. This is more expensive if you buy any significant number of upgrades, but that's the course you signed up for by implicitly restricting yourself to their install media. By mixing and matching, you get the negatives of above with respect to support (HP can blame the generic Promise chipped card, and vice-versa), but you pay more for the privilege of support that is compromised by the choice.
I'm a professional linux guy working for a hardware vendor. We invest a lot of time and money in making sure all our hardware works well for given linux distributions. I occasionally have to work with a customer who ultimately admits to third party options in the systems that usually end up the cause of their problem for reasons more purely technical than artificial CD key barriers. I'm a little defensive of this circumstance because even without artificial key measures introduced, this strategy can screw you over regardless of your software platform.
There are community projects that do explicitly do Long Term Support, Ubuntu 6.06, CentOS are two glaring examples of making a point of it, and Debian in practice has been that sort of distro. Gentoo, Fedora, et. al are catering to a different, more aggressive sort of mission that somewhat conflicts with the notion of legacy support (frequent releases produce too many overlaps in a long term model, having to maintain 4-5 different trees is not feasible. Ubuntu is doing a decent compromise (6.6 is long term, and they plan to do that once in a while, but still have frequent, shortly maintained releases to acheive the better of both worlds).
Fedora's whole mission is pretty much implement the cutting edge and let people experiment and play with it. The target audience was inteded to be those desktop users/enthusiasts who would jump on the current release anyway, since it is free. If any business saw a distribution like Fedora and thought it a good idea to base their infrastructure on a Fedora Core release, they are now getting basically what they asked for.
I don't think this is necessarily a bad thing, it's more an acceptance of what people should have perceived all along, Fedora isn't trying to be that kind of distribution. If you want that kind, go to a commercial vendor (RHAT, Novell) or go with something like CentOS or Debian, which have clear missions/policies that align with that sort of usage. Could consider Ubuntu, but I wouldn't be that confident in Canonical yet, but the LTS is at least a stated mission and Canonical has a vested business interest in being considered a serious business worthy option, while Fedora obviously hs no such vested interest. Similarly, I wouldn't use Gentoo or OpenSuSE in those contexts either, their missions are valid, but not in line with common business desires/needs. Debian and CentOS do, and generally end up 'boring' in the eyes of enthusiasts, and Fedora Core, Gentoo, Ubuntu's 6 month releases, etc all are generally more interesting to the enthusiast, but can't provide legacy support and the cutting edge all the time.
I can't be so sure in doubting it. Once they finally, eventually, get the licensing into the right terms, things can look different. Vastly different. In at least 2 perspectives: 1. We still encounter the 'Who the fxxx is Linux !' on the corporate level. SUN is perceived differently. 2. With a bit of sun and traction, we'll get the blobs (IPW) into it rather easily and drivers as well. Well, first, on licensing, Sun is where they want to be there, sort of like OSX/OpenDarwin licensing, except Apple left the more boring underpinnings free and kept the cool GUI level stuff closed, while Sun effectively opened the cool stuff, and kept the shitty stuff closed (no CDE open source.... such a... shame...). Aside from GPL purists, the licensing they are using I don't think of as particularly an impediment to a lot of usage. Their license may be a bit more open to getting taken advantage of than GPL, but ultimately it's technically not much more vulnerable to acts than the LGPL (except LGPL requires dynamic linking for different licenses, so ldd will show it obviously, while CDDL allows different.o files to have different licenses, making it very hard for a user to know where the CDDL based stuff is vs. non-CDDL stuff). CDDL is basically the MPL, and firefox's success shows MPL is hardly an obstacle that can prevent acceptance.
On "Who the fxx is Linux !", it sounds like living in 2000 or earlier. Last I heard someone speak something along those lines from a business perspective was early 2003, as a small company president reluctantly started pulling linux support in because their customers demanded it. For some hold outs who just couldn't bring themselves to accept a new company like RedHat or SuSE, IBM's and HP's significant endorsements pushed most of those far over the edge. Even die-hard Sun people recognize that Sun flirted with heavier Linux efforts for a time before deciding to back off and give all they could give to out-do linux at their own game but with Sun as the obvious premiere vendor. Businesses now recognize what Linux as a market is and really like it. History has proven if something is business-advantageous in particular ways, it outweighs technical advantage to some extent.
Basically, if new computing technology does or could conceivably make or break a company relative to a competitor, it's dangerous to tie your success to one company without the option of switching when all the vendors aren't clearly different. You buy Solaris-Sparc, you have tied yourself hard to Sun's performance, it's expensive and time-consuming to switch, and your other vendors could leave you out to dry based on the platform. This is one of the major reasons Windows experienced such success in the business well before it was seriously ready for prime time. You bought an NT server from Dell, and down the road if Dell screws up or tries to screw you, just buy from HP, etc etc.
If you were a shop heavily invested in PA-RISC, HP screwed you when they said screw PA-RISC. This is exactly what a client hates the thought of. They recognize the benefits of Unix, but don't like being tied to a vendor.
Now the x86_64 platform with linux is the ultimate platform for this. Anywhere along the chain a vendor can make a mistake or go down and the clients are left with the option of a different vendor. Intel screws up, fine, AMD keeps rolling, and vice-versa. IBM disappoints you, buy HP next time. Novell makes a questionable deal with MS you don't like, you can go to RedHat.
OpenSolaris is an important step, but if you want support from both your hardware vendor and software vendor on OpenSolaris, you *have* to go with Sun right now. If Nexenta becomes another RedHat and if IBM, HP, and/or Dell accept both Sun's and at least one other distributor's version, *and* the driver/community support improves, maybe. But that scenario involves a *lot* of time and money to be spent that mostly would go just to benefit Sun's wishes, without sufficient tangible benefit for the rest of the world.
Hey, I have seen and played with OpenSolaris, so that makes at least two, just I'm less enthusiastic than he is... Though my short run with the distributions may not be long enough to fairly evaluate the community, but it doesn't stop me from trying...
That vision of Solaris kernel coupled with GNU userland, Xorg, traditional stuff found in a linux desktop is largely achieved by Nexenta. If you want a taste, I'd try Nexenta's offering, it really is interesting and left me with a much better taste in my mouth than Solaris Express. Though I'm not usually anal about the use of linux to refer to a distribution, if you change the kernel describing it as "Linux but with a different kernel" makes me roll my eyes... Solaris Express Community gives you a taste where Sun is going with it (what Solaris 11 will be like roughly), and Nexenta is a good picture of where the community would take it in a debian like way. Belenix (I didn't try, but my understanding is that it) is more ports-inspired in package management, so you cover three significantly different approaches with those distributions that fit various tastes needs (old-school sun, debian, and BSDs/gentoo). At least Nexenta seems to be trying to make a name for themselves as an alternate Solaris support company, but we'll see how that goes, for now the only serious vendor is Sun, which will be a problem for them making inroads against linux. They have some things to bring to the table, but is it enough to make the IT industry do major investments to follow them in spite of the current disadvantages technically and business-wise? I doubt it.
On Nexenta as a desktop platform, the problem in the execution is that the kernel matters more than a lot of people think at first glance in terms of providing the 'friendly' stuff. OpenSolaris (both Nevada and Nexenta) seems to be doing good in the media management (plug in disc, right things happen), but power management has a *LONG* way to go, as does wireless (drivers are there for the most popular, but their WPA supplicant port only includes support for the atheros driver so far, left my IPW in the cold). Also they currently lack drivers for certain lines of controller cards, and a whole host of other stuff that linux has.
Fair enough about the development code, but was playing with it largely because the support for hardware isn't close in Sol10U3 for my evaluation hardware, so I pretty much had to do Nevada. I won't complain too loudly about instability with Nevada and derivatives stability, but it was the option I could exercise, and it seemed stable, just lacking low level support to do nice things in a desktop context. Nevada build 54 was honestly much much similar to the Solaris I remember, which isn't that exciting (I got out of a Sun shop when Solaris 8 was current, it felt a lot like that but with Gnome, ZFS and the other stuff, which was not drastically changed from 7, 2.6, 2.5, etc..), while Nexenta seemed to capture the essence of the biggest features of Solaris with a familiar and well-integrated set of packages with a better package manager. I think if not for linux success, I fear that ZFS, DTrace and such wouldn't have seen the light of day, based on Sun's history of not fixing what wasn't obviously broken over the years.
Having said that, let me also add that Solaris10 is documented. Heavily. Coherently. Completely. HPUX and AIX are close, Linux isn't even an also-ran in the documentation realm. On this, I still have to continually reject. In some measure, maybe, and maybe every building block is documented in man pages, but practically speaking I can get information much more readily about any linux facility and how to use it in most common contexts. Trying to do the same in Solaris didn't get me far. This is a simple reality of the pervasive nature of the market, and undoubtedly if OpenSolaris were in the same position as Linux is, the tides would be turned, but to say there is insufficient linux documentation as it stands today is a sort of denial. Also, the major enterprise vendors (RH and Novell) offer good documentation, and if you are comparing Solaris 10, you need to compare against a shiplevel Enterprise distro.
On Zones, I don't think it is useless or would be unused, but I'm not sure that virtualization is that much less attractive as a solution to the problems. If running Solaris, I'd do Zones, but if not I don't think most would miss it. chroot is good enough for a lot of things, and for the well compartmentalized approaches, virtulization can do it, if not a tad heavy-handed...
Anyway, the commercial unix world hasn't changed much over the years, AIX, HP-UX, Irix and friends don't do much exciting (rather the respective companies seem to be shifting more to LinuX), and Sun is doing some interesting features (after flirting with Linux before heading back to push Solaris). ZFS in particular I think would have been difficult in the Linux community compared to a mostly closed shop. ZFS to achieve what it did had to cross a lot of barriers in terms of dealing with block devices, software raid, volume managment, etc. Linux I suspect would have problems determining how each layer in the traditional sense would handle each part if trying to do such a project from scratch. I'm not sure if OpenSolaris was 'linux' level successful and implemented across the board by a number of vendors if they could have smoothly deployed ZFS in the community. The market cares about more than the technical, so a single-vendor solution for business reasons will still be shunned in favor of multiple equivalent vendors (though only one may be used, the options for the future are nice).
If you couldn't already tell, the biggest thing that made me do the OpenSolaris trial is the thing I came away being most enthusiastic about, ZFS.
OpenSolaris hasn't been big enough to experience fragmentation yet, so you can't judge how well OpenSolaris will hold up at scale of development community, and from my work with it, for it to be anything close to edging out linux, it has a way to go in the day-to-day stuff that *needs* a larger community. When it has a large community and *still* doesn't get into debates and doesn't have forks, let me know. It's interesting to note that in spite of the well-publicized debates and forks which all pull from the same tree and get merged back in over time, linux does a pretty damned good job regardless of the underlying debates...
Look at how many half-baked virtualization models there are for Linux- At last look, there was the commercial VMWare (which does what it is intended damn well), the popular Xen (which also does a spectacular job), and the still infant one they folded in. The first two are far from half-balked.
SATA support took years to come up to snuff and it's still half-baked. Recent 2.6 kernels wouldn't even boot on some AM2 systems. I *know* from personal experience that OpenSolaris today can't possibly criticize another platform for inadequate hardware support. Half the hardware on the system I tried OpenSolaris on as of Nevada build 54 didn't have drivers yet, but all of it is supported in linux. SATA support is working fine on *thousands* of linux boxes I support, and just because some kernel not in a distribution didn't boot on some system, doesn't mean linux has poor support for AM2 platform systems.
Likewise for poor filesystems...Solaris ZFS, even if it can't change pool types (ie, you can't go from a pair of mirrored drives to a triplet of RAID-5 like drives) solves problems no other linux filesystem has. I will give that ZFS has solved problems no other *filesystem* has, but there are non-filesystem solutions that are less effective/efficient than the ZFS way, but still viable, more on that later.
Namely, it scrubs the disk, not just testing readability but correctness (via checksums), and regularly walks its own filesystem structure and metadata checking for inconsistencies. I like the concept, but uber-paranoia can be implemented elsewhere to be fair, and serious storage vendors are so paranoid at the low level that no error should propagate to that layer if you have a good provider. I have seen early level systems and crappy generics that have silent data corruption, but never ever have seen a shiplevel server from a tier one vendor or storage subsystem have silent data corruption.
XFS is interesting, but if you want to criticize can be ignored, ReiserFS is a fair thing to complain about, but I tend to ignore that. Now a fair comparison would not be ext3 to ZFS, but ext3 to UFS. UFS and ext3 both prove ten year old concepts aren't necessarily bad things.
OpenSolaris has Zones/BrandZ, ZFS, and DTrace which are interesting, though I think Zones/BrandZ and DTrace are sufficiently compelling to convert, and ZFS is certainly appealing, but admittedly harder and less efficient methodologies can be done so that ZFS isn't an absolute must-have (i.e. block-level approaches to snapshotting and software raid, hardlinks/rsync for snapshotting, volume managers that understand how to grow filesystems). ZFS does have high-layer checksumming even above doing all those things better than anyone else on the market today, but a good storage vendor is so paranoid at the lower levels that the check becomes redundant.
Specifically Nevada build 54 and Nexenta alpha 5. They have some interesting technologies (specificaly ZFS, which is interesting, and Zones, which is a bit lower overhead than virtualization, but not as flexible (everything still goes through the Solaris kernel)). Nexenta I honestly thought was a cool concept, and executed quite well, Debian package management and GNU software that is clearly better than some of the Sun basic utilities (and much less Java inclined...), offering the benefits that Solaris does have to offer...
Anyway, from my working with it, I know the OpenSolaris is certainly full of themselves, and some denial, but I don't think they can live up to their own expectations. For example, any complaint or bug frequently got met with 'at least you aren't running linux!'. They trashed on lack of documentation in linux while I struggled to find some documentation on their stuff that seemed unwritten. They'd pick up a decade-old howto and say 'this is how linux requires you do it, versus our not-yet released way, see how crappy linux is'. When people talked about how woefully (understandably) incomplete their ACPI and suspend support was, they pointed at linux and said 'linux acpi support hardly works at all, so don't expect too much' despite the reality of 3 out of 3 generic motherboards I've tried worked splendidly with linux acpi. My laptop despite being one of their officially tested still doesn't have clock modulation and their acpi parser barfs on the DSDT that nothing else (not even intel's compiler) even warns on. People discussing panics/hangs are met with 'at least it doesn't crash as much as linux', despite evidence to the contrary. They are used to a closed, proprietary world of a select set of hardware and the open world if they make any headway in is going to give them quite the wake up call. They talk about how much better their driver support is, despite the glaring lack of drivers. Largely their efforts in expanding that involve porting drivers from the BSD projects.
Anyway, their current implementation does admittedly seem adequate for most server type activities if the hardware is supported. I could see a lot of hardware vendors happy about a system with a stable binary interface for drivers that doesn't require rebuilds for every uname -r, but hardware vendors face the market realities and put up with the pain if they want to play in the server space. I understand the hassle, but linux making a PITA for hardware vendors have given us a lot more driver source than we could have hoped for. For the market, probably the single best card they have is ZFS. They have done a good job of consolidating volume management, software raid, filesystem, stuff like snapshots, and paranoia of checksumming everywhere into a single implementation. In doing so they have done things more efficiently (such as RAID format on disk leveraging filesystem layer knowledge for better performance), and trustworthy (a controller failing to report data corruption is detected at a higher level). ZFS is impressive, and that was/is the one thing that makes me really want the rest of the platform to be usable for me day to day.
DTrace is much hyped, and very useful in the hands of good developers and good administrators, but I don't see administrators at large making use of it enough to deliver on the hopes Sun sets up for it.
Zoning is a nice logical extension from simple chrooting which is more comprehensive, and more efficient than the other extreme of virtualization, theoretically. However, with virtualization being ubiquitous and most of the market accepting the ever-reducing overhead for the flexibility, I don't know if Zones are going to excite anyone that much. The BrandZ extension of the metaphor gives it some flexibility, but again their Linux profile still doesn't run linux things just right, and a linux vm with the linux kernel already will do so today.
So you have a platform that probably won't need to be as successful as linux had to be in order for hardware ve
In one thread... going... to make head.... explode....
Re:But universal close tag not flexible enough...
on
The NSFW HTML Attribute
·
· Score: 2, Informative
Ugh, fine, replace bold with a LI. LI can legally have P as a child, so a generic close tag followed only by a open LI and open P tag could refer to either. since an LI does not have to be closed (explicitly or implicitly) before P opens. You have a P nested in an LI and you ask to close something, the browser doesn't know without future context. You've changed the task of the parser to require look ahead rather than knowing what it is doing based on parsing done so far, and the parser has to evaluate things more carefully in the context of the HTML nesting rules. Add to that there may be cases I can't readily produce where a reasonable guess about which tag is being closed is impossible (in the LI, P case, if the close tag had anything other than LI, it would have to be a close P tag I think, since UL/OL can't have anything but LI nested underneath, and a new LI or close UL/OL would mean both closed regardless of which explicitly closed). The short of it is, HTML dictated closing tags that match opening tags, and HTML has never been explicitly designed since to handle the concept of generic closing, and some concepts exist that would break.
When I say do not require closing, it is clear I mean does not require explicit closing. implicit closing is a given.
Anyway, if anything, the examples show how much more complicated an HTML parser has to be to evaluate generic close tags.
Car alarm cost is something that buyer would have a possibility of voluntarily ponying up.
Anti-piracy measures are something a buyer would almost never voluntarily pay extra for given a choice.
Both have costs, but you can't directly link issues anti-piracy in software to protect the seller to something like anti-theft which is not controversial because it protects the consumer.
And removing anti-theft and key systems allows for cost reduction of a car somewhat, it's not a bizarre suggestion, but the difference would be a customer wouldn't *want* to not have locks and keys for a car.
Give up, your analogy is bad, it doesn't make the core concept easier to understand than the raw concept and it makes false comparisons. Analogies are *supposed* to bring something beyond understanding into a context understandable. Everyone knows exactly why MS wants to lock down their software, so you don't need an analogy to 'get it'.
WTF is up with slashdot (or submitter) calling 18-15 year olds the iPod generation, and wtf is up with CNN saying 'the web generation'? Both are stupid, and imply incorrect things. The implication seems to be that the iPod or the web destroyed interest in the space program. I may be out of that age range, but I can imagine if I were in an age group identified by the web or, even worse, a particular company's product, I'd be offended.
I was wrong, but still provided an example elsewhere in this thread that I think should be valid that is also ambigious. I went to http://webtips.dan.info/nesting.html to get straightened out. But still, with optional closing tags on some elements, things still get impractical.
In the car case, the features are often *optional* and the customer may want it and consciously pay more to protect their purchase. The car manufacturer/dealer from a business stance doesn't give a rats ass if that vehicle is stolen after purchase, and therefore doesn't force the alarm system down consumers throat, unless their market is overwhelmingly demanding the alarm system so that it is more expensive to maintain two options than it is just to include it across the line. Again, nothing to do with protecting the seller's interests, and everything to do with the customers. If a car is stolen, the customer is clearly victimized and the car company either doesn't care or even benefits (if customer eventually had to purchase another car).
In the software case, it is the manufacturer making it a bigger pain in the ass for the consumer *and* increasing costs passed to the consumer for the sake of protecting the *seller's* interests. If a customer has his license key duplicated or his software copied, the act in and of itself is no sweat off his back, just bad for the seller *if* the recipient of the data/key may have otherwised purchased it. If the recipient would not have purchased it no matter what, it could be considered bad (no direct harm, but setting an example for others who would buy), neutral, or good (larger customer base is good for a platform you want to sell developers on).
To use a more appropriate analogy, if someone bought a tape and duplicated it for a friend back in the day, no one thought to call it stealing. Now with the act abstracted so much through putting things up for anonymous download, the copyright holders are more successful in getting people to think 'theft', but the core reality of the situation hasn't changed in a way as to make the word 'theft' any more valid. It may be bad/infringing on rights, but theft is not the correct word, and any cost/inconvenience incurred by the customer is not for their own direct good as it is in the case of actual theft.
And in response to calling the C language use of } a universal close tag, that's not exactly right. {} just mean treat enclosed code kinda like a sigle statement. if, while, for, etc all take one statement.
if (a==b)
printf("a==b\n");
Is valid C code without any braces. Similarly:
{ a=0; b=2; c=b+a; }
without any conditional/loop/function declaration is also valid C code. It simply isn't the same kind of concepts as you see in a markup language. A universal close tag is *by definition* ambiguous, I don't see how you can say it eliminates ambiguity even if HTML only allowed things to nest instead of overlapping, you could have said it was simpler, but not less ambiguous.
There are instances where elements can be nested not in the order they are opened. For example, having an underlined and bolded sequence intersect would be such a case:
<i>this <b>is a</i> test sequence</b>
It seems silly, but it is valid html that doesn't perfectly nest as would be required for a universal close tag.
Yes, all those sites are highly popular, but there were sites that were popular then too. Just because you are popular doesn't mean your business plan is viable in terms of long term expense to income ratio. To some extent, companies then that may have otherwise been reasonable got caught up in the hype and overextended themselves. I worked at the time for a company that had been stable for several years before the excitement of the late 90s. They weren't particularly exciting, but they held their own. Their business plan was viable, though conservative. The whole dot-com thing started rolling, getting formerly sane business people excited, they started getting ready to go public, bought expensive computer and networking equipment because they thought it would help, started dishing out perks to employees and overpaying for consulting, etc etc, all on the premise that going public in the dot-com scenario and their product would be perfect. Needless to say, that company is effectively defunct now. Same here, I hate to be doubtful of everyone's current favorite company, but Google's situation stinks to high heavon of the.com days. They are a popular site that gives services for free and relies on advertising revenue. Google's advertising is more directed and popular, therefore worth more, but at the same time based on their recruiting and their ungodly number of servers, they are simultaneously pissing money away like crazy. Maybe it is sustainable, maybe it's only sustainable transiently (and advertisers will get bored, refuse to pay so much, etc), or maybe their expenses are exceeding their income until their hope of things turning out right or a plan of theirs is hatched. It's just an unprecedented amount of money being spent and infrastructure for an almost entirely ad-supported business. Business cases can matter, if a site is a business and their business case flops, they will not be able to provide the popular service.
Same with myspace and youtube (now google). They are immensely popular and sometimes usefel, but is the business case sane? You say the media has been buzzing about what youtube ultimately did for the last ten years, but buzz does not mean it isn't a bubble. Buzz in fact is a strong indicator that it could be a bubble.
Ebay and Amazon last I checked were doing just fine, with a healthy and obvious business case. These are examples of businesses that arose during dot-com days with staying power and a sound business case. Their existence did not mean there was no bubble since two companies could be pointed to as not failures.
The other sites, some are overhyped sites that will fail, some are not getting caught up in overextending themselves (Wikipedia is not a business, and from all accounts is not going spend-happy, so I think they'll stay sane.
Now the one counterpoint is broadcast television, also almost entirely ad supported, and probably if you put the investment nationwide into television broadcast stations, you probably are spending more than google. By the same token, however, the ads in broadcast television until recently have been in the users face and not practically skippable (VCRs too much a pain in the ass to record every show, DVRs make it painless), and in the face of DVR installations, many primetime shows have resorted to realtime giveaways/contests to encourage people to watch before their DVRs can hope to have enough recording to skip commercials.
Oh wait, they did that too with red, green and blue LEDs, which is killer because there's no gold at the end of that rainbow effect The fundamental cause of 'rainbow effect' is not a wheel spinnig around, it's having only one primary on the screen at any given instant in time. The LED array if implemented properly will have the equivalent effect of very high RPM color wheels (which they already are doing). I haven't noticed rainbox effect, but if someone theoretically did notice with high rpm wheels, they might theoretically notice with LED array. I don't know how they time the LED array intervals and by extension what the equivalent RPM/size of a color wheel wold have to be to equal it, but at the end it's the same basic strategy.
The LEDs definitely have shorter time to full brightness, I hear they enable the sets to do a richer color gamut, and of course are much more longer lasting and less sensitive to power cycles. With LEDs DLPs ought be *the* sets to have if you have longevity in mind. LCD panels theoretically could break down more easily under heat, and I know damn well how easily LCD panels get dead pixels (every flat panel I've had developed at least one screwed up pixel within two years). The heat doesn't apply to flat-panel sets so much, but the dead pixels do. Plasma I hear has been criticized for losing quality over years, as well as burn-in. There have been improvements, but still... You're left with LCoS type systems, CRT, and DLP. DLP and LCoS may benefit equally from LED light instead of lamp. CRT projection has been very solid over the years, but subject to burn in.
One, most single-chip DLP units nowadays have the color wheel spin at such a high velocity as to render the effect impossible to perceive. My color wheel spins about three times as fast as color wheels during the time people complained some could notice the transitioning.
Second, on more expensive DLP they have multi-chip units, with a chip per color and no wheel. You are back to worrying about convergence, but rainbow effect is impossible.
Third, newer sets are starting to employ high-output LED arrays instead of lamps. This means that for one lower power and heat as well as longer life than bulbs (not to mention cutting out the need for some moving parts), and for another they can flash the different primaries very quickly to the chip. Of course, if they don't flash quickly enough, it could still happen.
Slashdot Mirror
Milton is that you? Did you get your stapler back?
Use buzzwords such that whether they are lying or not is undefined...
One thing: its patsies, HP In this case, HP is not just being MS' patsy, they are serving their own interest. MS didn't request them cripple their customization, they crippled it themselves to encourage any upgrades to be HP supplied, not third party.
The key sentence where everything went to crap: Out came the anaemic 40GB drive from one HP, and in when the Promise controller and two WD 200GB SATA drives. In the first part, he drank the vendor kool-aid and got their customized XP install. That's a valid choice in and of itself. Then he grabbed a third-party controller card and expected the stuff HP provided to play nice with it. HP didn't want him to do that, they wanted him to buy a presumably much more expensive HP branded controller to do the same thing. HP's install CD not accomodating that is hardly a surprise, and hardly a MS decision. The different keys for retail and OEM reflect the different pricing tiers. .
If they are a particularly small business, not going with one vendor is a valid choice, but you best put it together via all-third-party parts and get a generic OEM windows disk. If you can get a no-windows discount on the HP system, and use that discount for a different license, you can go with a non-restricted install media set. You do, however, in this way accept a higher degree of risk (problem determination falls squarely on your shoulders, and your vendors may disagree with your conclusion and blame other parts..). If you run on thin margins and time is not uber-critical for systems, this may be the appropriate path
If you drink the vendor kool-aid and get their hardware and software, you've drunk the kool-aid and as a consequence, you ought buy from HP your upgrades. You can't expect something put together by them to work for hardware configurations they would explicitly not support. This is more expensive if you buy any significant number of upgrades, but that's the course you signed up for by implicitly restricting yourself to their install media. By mixing and matching, you get the negatives of above with respect to support (HP can blame the generic Promise chipped card, and vice-versa), but you pay more for the privilege of support that is compromised by the choice.
I'm a professional linux guy working for a hardware vendor. We invest a lot of time and money in making sure all our hardware works well for given linux distributions. I occasionally have to work with a customer who ultimately admits to third party options in the systems that usually end up the cause of their problem for reasons more purely technical than artificial CD key barriers. I'm a little defensive of this circumstance because even without artificial key measures introduced, this strategy can screw you over regardless of your software platform.
There are community projects that do explicitly do Long Term Support, Ubuntu 6.06, CentOS are two glaring examples of making a point of it, and Debian in practice has been that sort of distro. Gentoo, Fedora, et. al are catering to a different, more aggressive sort of mission that somewhat conflicts with the notion of legacy support (frequent releases produce too many overlaps in a long term model, having to maintain 4-5 different trees is not feasible. Ubuntu is doing a decent compromise (6.6 is long term, and they plan to do that once in a while, but still have frequent, shortly maintained releases to acheive the better of both worlds).
Fedora's whole mission is pretty much implement the cutting edge and let people experiment and play with it. The target audience was inteded to be those desktop users/enthusiasts who would jump on the current release anyway, since it is free. If any business saw a distribution like Fedora and thought it a good idea to base their infrastructure on a Fedora Core release, they are now getting basically what they asked for.
I don't think this is necessarily a bad thing, it's more an acceptance of what people should have perceived all along, Fedora isn't trying to be that kind of distribution. If you want that kind, go to a commercial vendor (RHAT, Novell) or go with something like CentOS or Debian, which have clear missions/policies that align with that sort of usage. Could consider Ubuntu, but I wouldn't be that confident in Canonical yet, but the LTS is at least a stated mission and Canonical has a vested business interest in being considered a serious business worthy option, while Fedora obviously hs no such vested interest. Similarly, I wouldn't use Gentoo or OpenSuSE in those contexts either, their missions are valid, but not in line with common business desires/needs. Debian and CentOS do, and generally end up 'boring' in the eyes of enthusiasts, and Fedora Core, Gentoo, Ubuntu's 6 month releases, etc all are generally more interesting to the enthusiast, but can't provide legacy support and the cutting edge all the time.
1. We still encounter the 'Who the fxxx is Linux !' on the corporate level. SUN is perceived differently.
2. With a bit of sun and traction, we'll get the blobs (IPW) into it rather easily and drivers as well. Well, first, on licensing, Sun is where they want to be there, sort of like OSX/OpenDarwin licensing, except Apple left the more boring underpinnings free and kept the cool GUI level stuff closed, while Sun effectively opened the cool stuff, and kept the shitty stuff closed (no CDE open source.... such a... shame...). Aside from GPL purists, the licensing they are using I don't think of as particularly an impediment to a lot of usage. Their license may be a bit more open to getting taken advantage of than GPL, but ultimately it's technically not much more vulnerable to acts than the LGPL (except LGPL requires dynamic linking for different licenses, so ldd will show it obviously, while CDDL allows different
On "Who the fxx is Linux !", it sounds like living in 2000 or earlier. Last I heard someone speak something along those lines from a business perspective was early 2003, as a small company president reluctantly started pulling linux support in because their customers demanded it. For some hold outs who just couldn't bring themselves to accept a new company like RedHat or SuSE, IBM's and HP's significant endorsements pushed most of those far over the edge. Even die-hard Sun people recognize that Sun flirted with heavier Linux efforts for a time before deciding to back off and give all they could give to out-do linux at their own game but with Sun as the obvious premiere vendor. Businesses now recognize what Linux as a market is and really like it. History has proven if something is business-advantageous in particular ways, it outweighs technical advantage to some extent.
Basically, if new computing technology does or could conceivably make or break a company relative to a competitor, it's dangerous to tie your success to one company without the option of switching when all the vendors aren't clearly different. You buy Solaris-Sparc, you have tied yourself hard to Sun's performance, it's expensive and time-consuming to switch, and your other vendors could leave you out to dry based on the platform. This is one of the major reasons Windows experienced such success in the business well before it was seriously ready for prime time. You bought an NT server from Dell, and down the road if Dell screws up or tries to screw you, just buy from HP, etc etc.
If you were a shop heavily invested in PA-RISC, HP screwed you when they said screw PA-RISC. This is exactly what a client hates the thought of. They recognize the benefits of Unix, but don't like being tied to a vendor.
Now the x86_64 platform with linux is the ultimate platform for this. Anywhere along the chain a vendor can make a mistake or go down and the clients are left with the option of a different vendor. Intel screws up, fine, AMD keeps rolling, and vice-versa. IBM disappoints you, buy HP next time. Novell makes a questionable deal with MS you don't like, you can go to RedHat.
OpenSolaris is an important step, but if you want support from both your hardware vendor and software vendor on OpenSolaris, you *have* to go with Sun right now. If Nexenta becomes another RedHat and if IBM, HP, and/or Dell accept both Sun's and at least one other distributor's version, *and* the driver/community support improves, maybe. But that scenario involves a *lot* of time and money to be spent that mostly would go just to benefit Sun's wishes, without sufficient tangible benefit for the rest of the world.
I meant to say "I don't think Zones/BrandZ and DTrace are sufficiently compelling to convert".
Hey, I have seen and played with OpenSolaris, so that makes at least two, just I'm less enthusiastic than he is... Though my short run with the distributions may not be long enough to fairly evaluate the community, but it doesn't stop me from trying...
That vision of Solaris kernel coupled with GNU userland, Xorg, traditional stuff found in a linux desktop is largely achieved by Nexenta. If you want a taste, I'd try Nexenta's offering, it really is interesting and left me with a much better taste in my mouth than Solaris Express. Though I'm not usually anal about the use of linux to refer to a distribution, if you change the kernel describing it as "Linux but with a different kernel" makes me roll my eyes... Solaris Express Community gives you a taste where Sun is going with it (what Solaris 11 will be like roughly), and Nexenta is a good picture of where the community would take it in a debian like way. Belenix (I didn't try, but my understanding is that it) is more ports-inspired in package management, so you cover three significantly different approaches with those distributions that fit various tastes needs (old-school sun, debian, and BSDs/gentoo). At least Nexenta seems to be trying to make a name for themselves as an alternate Solaris support company, but we'll see how that goes, for now the only serious vendor is Sun, which will be a problem for them making inroads against linux. They have some things to bring to the table, but is it enough to make the IT industry do major investments to follow them in spite of the current disadvantages technically and business-wise? I doubt it.
On Nexenta as a desktop platform, the problem in the execution is that the kernel matters more than a lot of people think at first glance in terms of providing the 'friendly' stuff. OpenSolaris (both Nevada and Nexenta) seems to be doing good in the media management (plug in disc, right things happen), but power management has a *LONG* way to go, as does wireless (drivers are there for the most popular, but their WPA supplicant port only includes support for the atheros driver so far, left my IPW in the cold). Also they currently lack drivers for certain lines of controller cards, and a whole host of other stuff that linux has.
On Zones, I don't think it is useless or would be unused, but I'm not sure that virtualization is that much less attractive as a solution to the problems. If running Solaris, I'd do Zones, but if not I don't think most would miss it. chroot is good enough for a lot of things, and for the well compartmentalized approaches, virtulization can do it, if not a tad heavy-handed...
Anyway, the commercial unix world hasn't changed much over the years, AIX, HP-UX, Irix and friends don't do much exciting (rather the respective companies seem to be shifting more to LinuX), and Sun is doing some interesting features (after flirting with Linux before heading back to push Solaris). ZFS in particular I think would have been difficult in the Linux community compared to a mostly closed shop. ZFS to achieve what it did had to cross a lot of barriers in terms of dealing with block devices, software raid, volume managment, etc. Linux I suspect would have problems determining how each layer in the traditional sense would handle each part if trying to do such a project from scratch. I'm not sure if OpenSolaris was 'linux' level successful and implemented across the board by a number of vendors if they could have smoothly deployed ZFS in the community. The market cares about more than the technical, so a single-vendor solution for business reasons will still be shunned in favor of multiple equivalent vendors (though only one may be used, the options for the future are nice).
If you couldn't already tell, the biggest thing that made me do the OpenSolaris trial is the thing I came away being most enthusiastic about, ZFS.
XFS is interesting, but if you want to criticize can be ignored, ReiserFS is a fair thing to complain about, but I tend to ignore that. Now a fair comparison would not be ext3 to ZFS, but ext3 to UFS. UFS and ext3 both prove ten year old concepts aren't necessarily bad things.
OpenSolaris has Zones/BrandZ, ZFS, and DTrace which are interesting, though I think Zones/BrandZ and DTrace are sufficiently compelling to convert, and ZFS is certainly appealing, but admittedly harder and less efficient methodologies can be done so that ZFS isn't an absolute must-have (i.e. block-level approaches to snapshotting and software raid, hardlinks/rsync for snapshotting, volume managers that understand how to grow filesystems). ZFS does have high-layer checksumming even above doing all those things better than anyone else on the market today, but a good storage vendor is so paranoid at the lower levels that the check becomes redundant.
Specifically Nevada build 54 and Nexenta alpha 5. They have some interesting technologies (specificaly ZFS, which is interesting, and Zones, which is a bit lower overhead than virtualization, but not as flexible (everything still goes through the Solaris kernel)). Nexenta I honestly thought was a cool concept, and executed quite well, Debian package management and GNU software that is clearly better than some of the Sun basic utilities (and much less Java inclined...), offering the benefits that Solaris does have to offer...
Anyway, from my working with it, I know the OpenSolaris is certainly full of themselves, and some denial, but I don't think they can live up to their own expectations. For example, any complaint or bug frequently got met with 'at least you aren't running linux!'. They trashed on lack of documentation in linux while I struggled to find some documentation on their stuff that seemed unwritten. They'd pick up a decade-old howto and say 'this is how linux requires you do it, versus our not-yet released way, see how crappy linux is'. When people talked about how woefully (understandably) incomplete their ACPI and suspend support was, they pointed at linux and said 'linux acpi support hardly works at all, so don't expect too much' despite the reality of 3 out of 3 generic motherboards I've tried worked splendidly with linux acpi. My laptop despite being one of their officially tested still doesn't have clock modulation and their acpi parser barfs on the DSDT that nothing else (not even intel's compiler) even warns on. People discussing panics/hangs are met with 'at least it doesn't crash as much as linux', despite evidence to the contrary. They are used to a closed, proprietary world of a select set of hardware and the open world if they make any headway in is going to give them quite the wake up call. They talk about how much better their driver support is, despite the glaring lack of drivers. Largely their efforts in expanding that involve porting drivers from the BSD projects.
Anyway, their current implementation does admittedly seem adequate for most server type activities if the hardware is supported. I could see a lot of hardware vendors happy about a system with a stable binary interface for drivers that doesn't require rebuilds for every uname -r, but hardware vendors face the market realities and put up with the pain if they want to play in the server space. I understand the hassle, but linux making a PITA for hardware vendors have given us a lot more driver source than we could have hoped for. For the market, probably the single best card they have is ZFS. They have done a good job of consolidating volume management, software raid, filesystem, stuff like snapshots, and paranoia of checksumming everywhere into a single implementation. In doing so they have done things more efficiently (such as RAID format on disk leveraging filesystem layer knowledge for better performance), and trustworthy (a controller failing to report data corruption is detected at a higher level). ZFS is impressive, and that was/is the one thing that makes me really want the rest of the platform to be usable for me day to day.
DTrace is much hyped, and very useful in the hands of good developers and good administrators, but I don't see administrators at large making use of it enough to deliver on the hopes Sun sets up for it.
Zoning is a nice logical extension from simple chrooting which is more comprehensive, and more efficient than the other extreme of virtualization, theoretically. However, with virtualization being ubiquitous and most of the market accepting the ever-reducing overhead for the flexibility, I don't know if Zones are going to excite anyone that much. The BrandZ extension of the metaphor gives it some flexibility, but again their Linux profile still doesn't run linux things just right, and a linux vm with the linux kernel already will do so today.
So you have a platform that probably won't need to be as successful as linux had to be in order for hardware ve
In one thread... going... to make head.... explode....
Ugh, fine, replace bold with a LI. LI can legally have P as a child, so a generic close tag followed only by a open LI and open P tag could refer to either. since an LI does not have to be closed (explicitly or implicitly) before P opens. You have a P nested in an LI and you ask to close something, the browser doesn't know without future context. You've changed the task of the parser to require look ahead rather than knowing what it is doing based on parsing done so far, and the parser has to evaluate things more carefully in the context of the HTML nesting rules. Add to that there may be cases I can't readily produce where a reasonable guess about which tag is being closed is impossible (in the LI, P case, if the close tag had anything other than LI, it would have to be a close P tag I think, since UL/OL can't have anything but LI nested underneath, and a new LI or close UL/OL would mean both closed regardless of which explicitly closed). The short of it is, HTML dictated closing tags that match opening tags, and HTML has never been explicitly designed since to handle the concept of generic closing, and some concepts exist that would break.
When I say do not require closing, it is clear I mean does not require explicit closing. implicit closing is a given.
Anyway, if anything, the examples show how much more complicated an HTML parser has to be to evaluate generic close tags.
Car alarm cost is something that buyer would have a possibility of voluntarily ponying up.
Anti-piracy measures are something a buyer would almost never voluntarily pay extra for given a choice.
Both have costs, but you can't directly link issues anti-piracy in software to protect the seller to something like anti-theft which is not controversial because it protects the consumer.
And removing anti-theft and key systems allows for cost reduction of a car somewhat, it's not a bizarre suggestion, but the difference would be a customer wouldn't *want* to not have locks and keys for a car.
Give up, your analogy is bad, it doesn't make the core concept easier to understand than the raw concept and it makes false comparisons. Analogies are *supposed* to bring something beyond understanding into a context understandable. Everyone knows exactly why MS wants to lock down their software, so you don't need an analogy to 'get it'.
WTF is up with slashdot (or submitter) calling 18-15 year olds the iPod generation, and wtf is up with CNN saying 'the web generation'? Both are stupid, and imply incorrect things. The implication seems to be that the iPod or the web destroyed interest in the space program. I may be out of that age range, but I can imagine if I were in an age group identified by the web or, even worse, a particular company's product, I'd be offended.
I was wrong, but still provided an example elsewhere in this thread that I think should be valid that is also ambigious. I went to http://webtips.dan.info/nesting.html to get straightened out. But still, with optional closing tags on some elements, things still get impractical.
Ok, so some research shows you are right, *however* there are tags that can, but do not require closing (because HTML is so fubar).
<TD>, <TR>, <P>, <LI>, can all either be closed or not, so how then, do you deal with the following segment of theoretical HTML?
some stuff to keep in mind <b>and here is the important part
<p>new paragraph that continues emphasis
</>
More text
This would be ambiguous and impossible to know whether the code intended to end the paragraph or stop bolding.
That's a really bizarre analogy.
In the car case, the features are often *optional* and the customer may want it and consciously pay more to protect their purchase. The car manufacturer/dealer from a business stance doesn't give a rats ass if that vehicle is stolen after purchase, and therefore doesn't force the alarm system down consumers throat, unless their market is overwhelmingly demanding the alarm system so that it is more expensive to maintain two options than it is just to include it across the line. Again, nothing to do with protecting the seller's interests, and everything to do with the customers. If a car is stolen, the customer is clearly victimized and the car company either doesn't care or even benefits (if customer eventually had to purchase another car).
In the software case, it is the manufacturer making it a bigger pain in the ass for the consumer *and* increasing costs passed to the consumer for the sake of protecting the *seller's* interests. If a customer has his license key duplicated or his software copied, the act in and of itself is no sweat off his back, just bad for the seller *if* the recipient of the data/key may have otherwised purchased it. If the recipient would not have purchased it no matter what, it could be considered bad (no direct harm, but setting an example for others who would buy), neutral, or good (larger customer base is good for a platform you want to sell developers on).
To use a more appropriate analogy, if someone bought a tape and duplicated it for a friend back in the day, no one thought to call it stealing. Now with the act abstracted so much through putting things up for anonymous download, the copyright holders are more successful in getting people to think 'theft', but the core reality of the situation hasn't changed in a way as to make the word 'theft' any more valid. It may be bad/infringing on rights, but theft is not the correct word, and any cost/inconvenience incurred by the customer is not for their own direct good as it is in the case of actual theft.
The phrase 'identity theft' doesn't make us cringe? I am about of sick of that phrasing as I was 'information superhighway' when it was popular...
And in response to calling the C language use of } a universal close tag, that's not exactly right. {} just mean treat enclosed code kinda like a sigle statement. if, while, for, etc all take one statement.
if (a==b)
printf("a==b\n");
Is valid C code without any braces. Similarly:
{
a=0;
b=2;
c=b+a;
}
without any conditional/loop/function declaration is also valid C code. It simply isn't the same kind of concepts as you see in a markup language. A universal close tag is *by definition* ambiguous, I don't see how you can say it eliminates ambiguity even if HTML only allowed things to nest instead of overlapping, you could have said it was simpler, but not less ambiguous.
There are instances where elements can be nested not in the order they are opened. For example, having an underlined and bolded sequence intersect would be such a case:
<i>this <b>is a</i> test sequence</b>
It seems silly, but it is valid html that doesn't perfectly nest as would be required for a universal close tag.
One, that analogy made zero sense.
Two, ad supported media businesses do have precedent, it's called broadcast TV.
Yes, all those sites are highly popular, but there were sites that were popular then too. Just because you are popular doesn't mean your business plan is viable in terms of long term expense to income ratio. To some extent, companies then that may have otherwise been reasonable got caught up in the hype and overextended themselves. I worked at the time for a company that had been stable for several years before the excitement of the late 90s. They weren't particularly exciting, but they held their own. Their business plan was viable, though conservative. The whole dot-com thing started rolling, getting formerly sane business people excited, they started getting ready to go public, bought expensive computer and networking equipment because they thought it would help, started dishing out perks to employees and overpaying for consulting, etc etc, all on the premise that going public in the dot-com scenario and their product would be perfect. Needless to say, that company is effectively defunct now. Same here, I hate to be doubtful of everyone's current favorite company, but Google's situation stinks to high heavon of the .com days. They are a popular site that gives services for free and relies on advertising revenue. Google's advertising is more directed and popular, therefore worth more, but at the same time based on their recruiting and their ungodly number of servers, they are simultaneously pissing money away like crazy. Maybe it is sustainable, maybe it's only sustainable transiently (and advertisers will get bored, refuse to pay so much, etc), or maybe their expenses are exceeding their income until their hope of things turning out right or a plan of theirs is hatched. It's just an unprecedented amount of money being spent and infrastructure for an almost entirely ad-supported business. Business cases can matter, if a site is a business and their business case flops, they will not be able to provide the popular service.
Same with myspace and youtube (now google). They are immensely popular and sometimes usefel, but is the business case sane? You say the media has been buzzing about what youtube ultimately did for the last ten years, but buzz does not mean it isn't a bubble. Buzz in fact is a strong indicator that it could be a bubble.
Ebay and Amazon last I checked were doing just fine, with a healthy and obvious business case. These are examples of businesses that arose during dot-com days with staying power and a sound business case. Their existence did not mean there was no bubble since two companies could be pointed to as not failures.
The other sites, some are overhyped sites that will fail, some are not getting caught up in overextending themselves (Wikipedia is not a business, and from all accounts is not going spend-happy, so I think they'll stay sane.
Now the one counterpoint is broadcast television, also almost entirely ad supported, and probably if you put the investment nationwide into television broadcast stations, you probably are spending more than google. By the same token, however, the ads in broadcast television until recently have been in the users face and not practically skippable (VCRs too much a pain in the ass to record every show, DVRs make it painless), and in the face of DVR installations, many primetime shows have resorted to realtime giveaways/contests to encourage people to watch before their DVRs can hope to have enough recording to skip commercials.
The LEDs definitely have shorter time to full brightness, I hear they enable the sets to do a richer color gamut, and of course are much more longer lasting and less sensitive to power cycles. With LEDs DLPs ought be *the* sets to have if you have longevity in mind. LCD panels theoretically could break down more easily under heat, and I know damn well how easily LCD panels get dead pixels (every flat panel I've had developed at least one screwed up pixel within two years). The heat doesn't apply to flat-panel sets so much, but the dead pixels do. Plasma I hear has been criticized for losing quality over years, as well as burn-in. There have been improvements, but still... You're left with LCoS type systems, CRT, and DLP. DLP and LCoS may benefit equally from LED light instead of lamp. CRT projection has been very solid over the years, but subject to burn in.
Three solutions commonly exist for that.
One, most single-chip DLP units nowadays have the color wheel spin at such a high velocity as to render the effect impossible to perceive. My color wheel spins about three times as fast as color wheels during the time people complained some could notice the transitioning.
Second, on more expensive DLP they have multi-chip units, with a chip per color and no wheel. You are back to worrying about convergence, but rainbow effect is impossible.
Third, newer sets are starting to employ high-output LED arrays instead of lamps. This means that for one lower power and heat as well as longer life than bulbs (not to mention cutting out the need for some moving parts), and for another they can flash the different primaries very quickly to the chip. Of course, if they don't flash quickly enough, it could still happen.