I'd be interested in how you do this, as I'm currently looking at a similar need for one of my clients and hadn't considered MMC via the SPI interface that we want to use in this project. That could solve the general unavailability problem with large SPI EEPROMs. You can reach me at: dubNO[at-sign]infoSPAMwave.com
It's interesting to note that the rap against NeWS was that it was glacially slow on 68000-vintage and first-generation SPARC chips. The world has come a long way in ten years, and NeWS would now scream on no more processing power than exists in many cellphones.
If Sun really wants to reinvigorate and solidify their rapidly vaporizing position on the technical desktop, they might consider a major refresh of NeWS to be released as open source software... Of course, that's what Java might have been had anyone paid attention when I pointed out that it needed two vital things: 1) a known, predictable place for the JVM (classpath doesn't cut it), and 2) some way to tell what versions of the JVM are available and some way to select a partciular one, if your app needs it.
Sure enough, these are some of the biggest problems today in Java - it's not the hard stuff, but the simple, stupid stuff that can sink a great idea. Always has been, always will be. (But it's hard to convince the CS PhD's that "real world" practicality matters more than their fixation on "elegance"...
* A way to send less data over the wire for toolkits such as QT / GTK+.
* Easier configuration and setup.
* Pluggable cut-and-paste architecture that can be more easily used by the other toolkits.
* Better video drivers*.
* I know... we're out of luck here without help from the video card manufacturers.
You forgot important things like a audio model and perhaps some way to do useful multimedia things like synchronize audio and video. X is both good and horrible, one thing it's not and may never be, is a multimedia platform. Over all, though, it's been needing to be replaced for over 10 years now, when I wrote up a paper at a former employer urging exactly that. In many ways XFree's success in finally making an X that (sort of) works has made life far more difficult for the much-needed alternatives. Keep the great ideas from X, like display abstraction, but please, please, re-do the rest...
Not that I'm holding my breath - I gave up and decided to run Windows on the desktop in 1999, despite being a dyed-in-the-wool Unix guy for nearly 20 years. So far, I've had no reason to regret it, and even recent tries with the new RedHat and Mandrake distros only reaffirmed that Linux (and BSD, which is better, but suffers from the same X problem) cannot really compete as a desktop OS on the same level as Windows and OS X.
It's a shame, because I'd love to see a real alternative, but there is none on the horizon, so I and countless thousands of others hold our noses and keep running MS, at least until all our apps stop working on 98SE and W2K, since I won't upgrade to XP...) At that point, I probably have no choice but to buy Apple.
Employee-ownership - but are they competent?
on
Inside SAIC
·
· Score: 1
I've always had a really bad opinion of SAIC since I ran into them as my first defense subcontractor at my first job with an aerospace manufacturer. Maybe it's not fair to carry a negative opinion after so long (It's been 18 years), but this was so bad that I've never forgotten it.
The setting is a now-merged aerospace company in Southern California. We had a contract from the Air Force to develop one of the stupidest devices I've ever run across - a "flexible" assembly fixture that could programmatically reposition all the "hard points" to support aluminum panels as they are riveted together. Granted, the fixtures aren't cheap - most of them cost at least several thousand dollars, but this was replacing a $10,000 system with a $5 million one. (Oh, and there are dozens in use on any given day in a single plant - the logistics is a pain, but get real...) Anyway, as a young robotics engineer, it seemed like an interesting project, even if it made no economic sense.
SAIC was contracted to do the basic design and preliminary feasibility analysis. (Why we didn't do it ourselves given their demonstrated incompetence, is beyond me...) I still remember how in no fewer than probably a dozen places in the document and presentation, their "brilliant" young engineer (who was actually a couple of years older than I was at the time) kept pointing out how absolutely vital it was that SAIC's design used six bolts (all into the same flat plate) to restrain the device in six degrees of freedom. He really believed you needed one bolt per degree to hold something to a flat plate! The document had been reviewed by SAIC's "more experienced" hands, but when we complained about getting crap for our money, they pretty much said, "if you don't like our design, do it yourself". We did, and I've never been willing to work with SAIC since then. (Unfortunately, on of the worst PHB's I've ever encountered in the real world forced a decision to keep far too much of the SAIC design he had spent so much of the company's money on, so the project was doomed. I transferred to another group, and never looked back...)
It's seen an unexpectedly wide adoption since 1983? If it takes that long to get unexpected adoption. how long does a slow rollout take?
No, that's 1993, not 1983. It only seems like twenty years.
My own suspicion is that IPv6 will take off just after the 12th of Never... Seriously, unless the ISP's start to support it, and there is some real end-user benefit (mobility, security, etc.), there's just no valid reason to even think about absorbing the substantial pain of an IPv6 transition. (Face it, it's all most IT organizations can do to get things working with v4. Does anyone really think those folks are going to switch to v6 unless they *have* to? Fat chance.)
After reading your list, it seems that perhaps the issues you have are with IP Masquerading, a special form of NAT where the external addresses are non-exclusive.
This is a great point - I often forget that most people, especially the Linux types hanging around here, have absolutely no idea what real NAT looks like.
They think that Linux' brain-dead IP-masquerading is the same thing as real NAT. It's not. It's not even close.
IP-Masq has lived far too long, and it's time to drive a stake through the heart of that vampire sucking the lifeblood out of our networks. Why no one has bothered to make a real NAT part of mainstream Linux distros and we still suffer on with IP-Masq, I really don't know. Ignorance, I guess...
I thought that everyone had either gone forward to style sheets or back to tables.
Actually, frames are more useful today than ever: In today's world where web pages are more like application interfaces than information displays, frames are a great way to avoid reloading things, because they encapsulate the frame content as a separate URL. DOM/CSS/JavaScript/XSLT and the like will make frames more, not less common. Check some of the more davanced uses of these technologies, and you'll find them wrapped up in frames all over the place...
Fortunately, given the very strong presence of prior art in this case (most especially the Netscape 2.0 browser itself) it seems that there's no reason not to use them. And if they are truly patent protected, it seems to me that if they don't restrict their use in the browsers, they can't reasonably expect to restrict them on the server end... SBC is just up to their usual moneygrubbing games. I'm fully in support of patents, but this abuse is enough to put "Call Birch to change phone service" on my to-do list for later this week...
"You're on an island, after a plane crash in which you were the only survivor. One one side of the island is a village full of maneating cannibals who are starving to death. In the middle of the island are wild boars, jungle cats, and all sorts of poisonous creatures. On your side of the island, there's nothing but rocks and the contents of your survival kit: one roll of toilet paper, some twine, a flaregun with three flares, a can of sardines, and a swiss army knife. What do you do first?"
Actually, the right answer to this one quite clear-cut: Respond with a couple of other questions, based on the following "Rule of threes" for survival (many people start by looking for food, which is a good way to die, since it's actually a fairly low priority from a life-preservation point of view):
You can live:
3 weeks without food 3 days without water 3 hours without shelter 3 minutes without air 3 seconds without thinking
Since I've got a can of sardines and some possibility of dining on wild boars, jungle cats, poisonous (hopefully just venomous, and not actually poisonous...) creatures, I've actually got quite a while before I have to start seriously worrying about food.
Assuming they haven't left out crucial information like I'm being crushed under the wreckage of the plane and can't breathe, a reasonable first question might be, then: "What's the weather like?" ("tropical paradise" may well make shelter a no-op), followed by, "Where is the nearest fresh water?" (And perhaps, "Do I have a sheet of plastic in my survival kit to build a solar still?") It's likely in this scenario that water is the number one priority.
Finally, it's important to recognize that things aren't *too* bad, if you've still got luxury items like toilet paper...;-)
While not wanting to turn this into a license fre-for-all, I'll just point out that I think the BSD license is generally superior, but would favor an LGPL derivative (sans the socialist Preamble) under one condition: That the author be able to distribute WITHOUT source code for a period of two or three years, after which the source must be made available. This seems, to me, to strike the best possible balance between commercial expoitation (which is realistically required to drive some significant advancements) and the common good. It's important that there be some period of exclusivity if we expect people to make non-trivial inventments in technology development.
(In one sense, this points to the root problem that patents last for an insanely long time in fast moving technology areas. I've proposed making the term of patents inversely proportional to the number of patents *granted* (not applied for) in that category on a sliding scale from two years at the low end to the current maximum for very slow-moving areas. This mechanism has the advantage of being self-regulating, and extending or contracting patent terms as activity heats up in certain areas.)
Mark my words, for the computing paradigms we have today, we've pretty much hit our limits. Don't hold your breath, any real improvements are 20 years out.
And we might as well just shut down the Patent Office, too, because everything that can be invented has already been done?:-)
Seriously, I think you're *way* off base here. The technologies to build a kick-butt distributed/networked filesystem have existed for years, but there has been no economic incentive to bring them all together. AFS/XFS/Coda/etc. show that there are a lot of interesting avenues for such development, but this is an area where open source may not be able to lead in terms of wide-scale adoption of such systems. On the other hand, if Microsoft, Sun, and say, Red Hat got togther to build an NFSng that all would put in their products, we'd have a *real* revolution.
The barriers to next-gen filesystems are primarily political and economic, not technical.
in a related study, engineering isn't necessarily the best way to be social. that jerk on the tour that told you chicks dig engineers was a lying bastard.
Actually, chicks do dig engineers. But scientists, especially computer scientists, are another matter, entirely...:-) [ducking behind fireproof object...]
Leap seconds are evil. As someone who has spent way too many hours programming high precision time distribution systems to deal with leap seconds, I'd say 'good riddance
Leap seconds may make life as a programmer more difficult (right now, I'm working on figuring out how long-term instrumentation systems should handle thier timestamps when leap seconds occur), but they are absolutely necessary:
The basic problem is that two things have to happen:
Calendars must be kept in sync so that any particular date in the year always occurs with the Sun in the exact same position in the ecliptic. (All navigation assumes the Sun rotates around the earth. Einstein came up with reletivity only after someone asked him to prove it doesn't, which, of course, he couldn't do...)
Clocks must be kept in sync so that noon occurs precisely when the Sun crosses the local meridian.
It's vitally important for navigation and many other purposes that our clocks be corrected as required to reflect the fact that the *real* units of time aren't some stupid atomic counter, but the positions of the heavenly bodies that are used to confirm all references. In essence, the real problem is that the lab rats don't want to recognize that their ridiculous wiggling atom setup is worse than useless in the real world, since it creates all kinds of synchronization problems and in any case, as Einstein said, we really have no more reason to believe the atomic clock is accurate than that the celestial clock is accurate. It's a relativistic problem - while it's probably more likely that the earth is slowing down, there's no way to prove cesium atoms aren't speeding up, oddly, while other research shows that the speed of light may be slowing down...:-)
Long live leap seconds - at least until we get around to redefining time based on celestial mechanics, which is the only thing that makes any sense in the real world, anyway...
The only replacement for AFS that is even close is Microsoft's "Win2k AD'd dfs", and even it is missing a large number of features that AFS has.
Most notably the ability to build a distributed filesystem that doesn't have a single point of failure, as was required by the MS DFS the last time I checked, since all the name re-mapping happens on a singel server. Ugh - what were they thinking? This has got to be one of the most egregious examples of Microsoft not thinking through the problem before writing the code.
On the other hand, MS DFS was never intended to be a real solution, but just to keep W2K from being disqualified because it didn't have heirarchical mounts. I've seen MS draw it like a gun in several companies where this was an issue - a situation that really sets me up to earn my consulting fees.;-)
Nothing else is - not NFS, not SMB, not whatever, just NCP.
You will want to look into AFS then before making such a judgement. It may not change your opinion, but for file serving there is very little lacking.
I first used AFS back when it cost a zillion dollars from Transarc and you had to deploy the whole stinking Athena environment to even use it. (This was years before IBM bought Transarc, but the prices are still ridiculous...) I looked at it hard again last year. It's interesting, and certainly improved, but I'm still not sure it's ready for prime time, especially since there are big chunks of functionality and platform support that are simply MIA.
I really don't hink it's possible for any third party network filesystem to have a major impact these days, since unless it's embedded in the major OSes, it's DOA. If Sun and Microsoft could work togehter for once, we might finally get a decent network filesystem for the 21st century, one with things a good distributed lock manager and fine-grained permissions. Oh, well, I can dream, anyway...:-)
They sold almost nothing, so in that sense they failed miserably. What they *did* do before Red Hat did it, was make it possible to say "Linux" in a meeting with a CIO and be taken seriously. They deserve some credit for that, even if reality didn't match the perception...
Isn't that what I said? (Poor grammar and ambiguous modifiers make following the thread quite difficult. In any case, my point was that BSD, MIT, and X licenses are not only open source (or Open Source, if you prefer), but also truly free, in that they are permissable for commercial use. This is the real difference between the BSD and GPL license camps: the GPL folks are primarily concerned with promoting a political agenda, while the BSD folks just want good code to be usable everywhere, recognizing that a rising tide lifts all boats...
Not exactly, it's a dumb terminal. It is basically a full-featured WinCE powered system with the sole purpose of mirroring what is on the servers display.
Think of it as doing a remote X display, if you are familiar with X11. Most of the gripes about it come with the first run of a new technology (from Microsofts point of view.) I would certainly love something like this that operates using X instead.
Actually, it has one major advantage over a Tablet PC - since it's not really one, and doesn't have to conform to Microsoft's ridiculous hardware profile in order to even be eligible to buy the TabletPC version of XP, it has a *real* touchscreen - one you can touch with your finger instead of a special active pen. In my mind, it's this requirement for an active pen digitizer that is the single stupidest thing abou the Tablet PC hardware profile, and may be a large part of the reason why this platform will not do as well as it should. (People want tablets, and the Tablet PC is the best yet, but still not good enough for a lot of folks...)
BTW - you wouldn't really want one of these that does X - that would be much slower than RDP, which is that rarest of creatures - a Microsoft protocol that's best-in-class.
Another thing that Linux folks should remember, but most of them couldn't spell Linux at the time
Insulting your audience on the first line is rarely a good way to get your point across.:)
It's really not an insult - there's more than a little truth to it - some of the people so ardently reciting the Linux party line here really *couldn't* spell Linux back then. The Linux community is great at creating converts, but very bad at giving them any historical context, or even acknowledging that (gasp) Linux may not be the world's best OS in some respects. I just thought that was kind of amusing...
You say in one bullet that Caldera was the first serious commercial linux, and in the next bullet talk about Red Hat preceding them. By the time I played with Caldera 1.0, Red Hat was pretty well established. Red Hat shipped their first version in '94, Caldera formed in '94 and didn't ship until some time after.
Red Hat was first, IIRC, but I stand by Caldera being the first *serious* commercial Linux. Red Hat was an early integrated distro, but they weren't viewed (and perhaps didn't yet view themselves) as a viable Linux vendor for the enterprise early on. Unfortunately, Caldera never articulated a really viable business model, and it sickens me to see them abandoning that work in favor of SCO.
Oh well, thank Bill Joy (and Theo de Raadt) for BSD!
eDirectory for Linux has been around for a while, I'll give you that, and Novell has a long (and sad) history with *nix. I'm not sure that the history of Novell and *nix is something to be proud of, however. That would be like talking about how qualified Novell is to contribute to the future of word-processors.:)
No question, there's been some exceptionally bad management and really bone-headed decision-making at Novell in the past decade. They certainly should have done much better given the ooprtunities they've had. (I think largely that's because they've never come to terms with the need to move to a Unix/Linux style OS as the preferred product platfrom, so even though part of the company recognized they should change, another part was too comfortable to allow it.)
Mainly, though, I just wanted to point out that Linux and Novell are more connected than many people may think at first...
Another thing that Linux folks should remember, but most of them couldn't spell Linux at the time:
Novell played a pivotal role in making Linux an option to be watched and considered. For those of you that don't know, so long ago that it's hard to find documentation of it on the web, Novell's then-CEO, Ray Noorda, had a skunk works project within Novell to build a revolutionary OS built on a strange free operating system called Linux. (IIRC, it was code-named "Corsair")
This fundamentally changed the course of Linux forever:
Corsair *was* revolutionary - a modern desktop OS with the strength of Unix(Linux) and another feature that had never been integrated, or even bundled with the OS at that point: a web browser. This was big, visionary medicine in 1994!
That little skunkworks project was given a fair amount of press coverage, and for the first time, the concept of Linux was widely spread in the PC newspapers and magazines of the day. The fact that this was a pet project of Noorda's only added fuel to the interest and speculation. Many of us who had been looking to promote Unix/Linux systems used this buzz as a door-opener in many organizations. (Server versions with much of NetWare's functionality were expected to result too - but it seems we're just now getting those...
When Noorda retired, Bryan Sparks and Ransome Love approached him to fund Caldera, the first serious commercial Linux company, giving the OS another shot in the arm and yet more credibility as possibly having more potential than the average Finnish student's hobby. Red Hat was very small, and not really taken seriously by anyone other than a few hobbyists, at least partly due to one of the goofiest names to hit the industry since that wave of fruits in the 1980's.
Caldera in many ways is probably the most important and influential Linux company ever. Even Red Hat, which preceded them, I think, owes much of what theier product is today to Caldera. For one thing, the Caldera folks were the first to recognize that Linux needed an installer. Caldera had the first Linux installer, the first graphical installer, and the first integrated text and graphical system administration tools. (LISA is still better at some things than mmost of what we live with today...) We old-timers remember when Linux was a little more than "some assembly required" - even if you knew what you were doing (and almost no one did, as fast as things were changing back then), getting everything assembled, installed, running, and properly configured could take dozens of hours. Caldera was the first to make a serious attempt to make Linux accessible to non-masochists. They deserve a lot of credit for that.
Finally, it was Caldera that showed the world that open source software could offer good value: For the unheard-of price of $99, you got a Linux that was easy to install, included a graphical web browser environment, a serious OS, and a web server. Although it's hard to remember (now that Apache is everywhere) this set of capabilities often cost thousands of dollars back then.
Finally, Caldera very nearly hired that young Finnish student, Linus Torvalds, and history (and Slashdot) might have judged Caldera very differently had they done what it took to get him there. (He reportedly turned down a great offer because Caldera would not guarantee that Linus could remain purely on the technical side, as the University of Helsinki did.)
Anyway, it seemed to be worth pointing out to those that don't know or remember how Novell and Linux have been related for a decade now, and why it's very good news to see the two renewing their relationship. As Rick said to Captain Renault as they walked in to the mist in Casablanca, "Louie, this could be the beginning of a beautiful friendship..."
Most of what Novell does is rather mature on that level. Much more so then Linux, but probably not as much as he thinks. It has great directory, authentication and network file systems. A good AFS, LDAP, Kerberos run Linux domain is perhaps less of a polished product then Novell, but it is not far behind.
Sorry, but Linux isn't even in the same league when it comes to network services. NetWare has its warts, but so far as NOS capabilities they are in a class of one. (Although Banyan was interesting, are they still around?)
The fine granularity of file permissions in NetWare is an absolute dream, and matches and supports real-world needs *far* better than those of Windows, or especially Unix-derived servers. (I've been dealing with the brain-dead Unix file permissions for 18 years now, and the whole system is a major dog's breakfast.)
ACLs have been grafted onto various network filesystems in myriad incompatible, incomprehensible, and unmanageable ways, but that's really no substitute for a just having a reasonable set of permissions capabilities in the first place.
Further, NDS is far and away the best directory service available today - it's really a shame it hasn't taken hold in the Unix/Linux world, as we need it badly if there is ever to be any hope of holding AD at bay. (Those that don't do serious enterprise work fail to comprehend that it's AD that makes it virtually impossible to pull Windows out of an organization - this is the *real* Kool-Aid, and if your organization has drunk a long draft of it, you're poisoned, bucko... Raw LDAP is not really an option in most environments, as the staffing required to manage it that way exceeds the available talented labor pool in most places...)
You're right that all the apps built on these network services have no real equivalent at all in the Unix/Linux world, and only shabby imitators in the Windows world, but even at the server-only level, NetWare in unequalled. I hate the way you administer it, (it's intentionally obtuse to encourage CNE certification), and it has some weaknesses as an application server, but it works and works well.
It's well-engineered, too: As a protocol jock, I say that with real knowledge - compare the rock-solid reliability, wide area bandwidth efficiency, and latency insensitivity, not to mention advanced features and security of NCP to *anything* else, and I think it will come out *way* ahead. I've built worldwide remote site networks that have to have transparent file access back to civilization via a satellite telephone, (the worst latency environment within three planetary diameters) and only NetWare and NCP are capable of operating in such an environment. Nothing else is - not NFS, not SMB, not whatever, just NCP.
Gee, this sounds like a Novell ad - It's not, I haven't even touched the product in two years, but what they do, they do well.
I din't say dying, but it's not well. Clearly, Linux is going to remina in the embedded space for a long time, but it seems much more current effort on the part of the hardware vendors is going to getting things ready for CE rather than Linux.
I couldn't agree more that Lineo and Metrowerks couldn't justify thier price premiums - and that's kinda the point...
Just *try* and find a small lightweight wireless webpad-type device that's not running CE right now. So far, I count only one, and it's an obsolete product that is not very rugged and not being upgraded.
I don't like what I see, but the trend is undeniable.
This is a business admission that the market share for winCE is, in fact, dropping like a stone. The embedded market uses Linux, which allows the tailoring of capabilities, a general understanding and lots of third party documentation ( like o'Rielly).
Sorry, I'm in the middle of an embedded project requirements specification right now, and you're way off base. Take a look at companies that used to have a notable embeded Linux bias, and many are either rushing to CE (Intrinsyc, Tiqit) or have been rendered ineffective in other ways (i.e., Lineo/Embedix absorbed into the Metrowerks amoeba.)
Embedded Linux and/or BSD are attractive, but it's proven to be a very difficult market to make money in. Well over half the several dozen links I collected six months ago on embedded Linux SBCs either belong to companies that are now defunct, or to companies that are deliberately trying to distance themselves from the profitless Linux market. And these are the *hardware* guys - It's even worse amongst the software companies! (Come on, I bet 90+% of the people here would never consider paying for an embedded Linux distro - the perceived value is just not there, except for a very few customers that can't provide enough revenue to ensure survival.
This is a basic economics problem, and one that may not be resolved in favor of open source. (Although I will show my stripes and say that I think BSD is gaining in the embedded space precisely because it permits recovery of investments in ways that aren't do-able under the GPL. Well that and the fact that NetBSD runs on darn near *anything*...)
What MS are saying with this deal is "we are the only ones allowed to licence WinCE. You can add to it, change it, remove stuff, so long as you tell us. BUT a copy of WinCE is still ours to licence. Sounds fair to me.
This is good news, and it is a completely legitimate way to open source something. One of the author's principal rights that many open source licenses neglect is the right of the original creator to continue to direct the development of code, especially to keep it philosophically in line with the original intent.
Although, like all freedoms, it can be abused, this can be a very good thing, too. It is precisely this sort of control by Sun that has kept Java from fragmenting far worse than it has. Ditto for Netscape and Mozilla. (Remember Mozilla is dual-licensed, a concept that has not been proven valid in court, and may not, since it defines inherent contradictions.)
This is really what all squabbles about intellectual property boil down to: how much control does an author or creator have after the fact?
This announcement is very good news, although I fear it means that MS will now roll through the embedded world virtually unopposed. But on the other hand, if no one else is capable of mounting a defense, then perhaps they shouldn't. CE is architecturally ugly, but it works much better than it used to, and one of these days may have enough features of a real OS to be a contender. As I posted here just recently, all the embedded hardware OEMs (even the ones that started out carrying the Linux flag) are rushing to CE to survive, and *that* trend will only grow with this announcement.
Slashdot Mirror
I'd be interested in how you do this, as I'm currently looking at a similar need for one of my clients and hadn't considered MMC via the SPI interface that we want to use in this project. That could solve the general unavailability problem with large SPI EEPROMs. You can reach me at: dubNO[at-sign]infoSPAMwave.com
It's interesting to note that the rap against NeWS was that it was glacially slow on 68000-vintage and first-generation SPARC chips. The world has come a long way in ten years, and NeWS would now scream on no more processing power than exists in many cellphones.
If Sun really wants to reinvigorate and solidify their rapidly vaporizing position on the technical desktop, they might consider a major refresh of NeWS to be released as open source software... Of course, that's what Java might have been had anyone paid attention when I pointed out that it needed two vital things: 1) a known, predictable place for the JVM (classpath doesn't cut it), and 2) some way to tell what versions of the JVM are available and some way to select a partciular one, if your app needs it.
Sure enough, these are some of the biggest problems today in Java - it's not the hard stuff, but the simple, stupid stuff that can sink a great idea. Always has been, always will be. (But it's hard to convince the CS PhD's that "real world" practicality matters more than their fixation on "elegance"...
What X needs:
... we're out of luck here without help from the video card manufacturers.
* A way to send less data over the wire for toolkits such as QT / GTK+.
* Easier configuration and setup.
* Pluggable cut-and-paste architecture that can be more easily used by the other toolkits.
* Better video drivers*.
* I know
You forgot important things like a audio model and perhaps some way to do useful multimedia things like synchronize audio and video. X is both good and horrible, one thing it's not and may never be, is a multimedia platform. Over all, though, it's been needing to be replaced for over 10 years now, when I wrote up a paper at a former employer urging exactly that. In many ways XFree's success in finally making an X that (sort of) works has made life far more difficult for the much-needed alternatives. Keep the great ideas from X, like display abstraction, but please, please, re-do the rest...
Not that I'm holding my breath - I gave up and decided to run Windows on the desktop in 1999, despite being a dyed-in-the-wool Unix guy for nearly 20 years. So far, I've had no reason to regret it, and even recent tries with the new RedHat and Mandrake distros only reaffirmed that Linux (and BSD, which is better, but suffers from the same X problem) cannot really compete as a desktop OS on the same level as Windows and OS X.
It's a shame, because I'd love to see a real alternative, but there is none on the horizon, so I and countless thousands of others hold our noses and keep running MS, at least until all our apps stop working on 98SE and W2K, since I won't upgrade to XP...) At that point, I probably have no choice but to buy Apple.
I've always had a really bad opinion of SAIC since I ran into them as my first defense subcontractor at my first job with an aerospace manufacturer. Maybe it's not fair to carry a negative opinion after so long (It's been 18 years), but this was so bad that I've never forgotten it.
The setting is a now-merged aerospace company in Southern California. We had a contract from the Air Force to develop one of the stupidest devices I've ever run across - a "flexible" assembly fixture that could programmatically reposition all the "hard points" to support aluminum panels as they are riveted together. Granted, the fixtures aren't cheap - most of them cost at least several thousand dollars, but this was replacing a $10,000 system with a $5 million one. (Oh, and there are dozens in use on any given day in a single plant - the logistics is a pain, but get real...) Anyway, as a young robotics engineer, it seemed like an interesting project, even if it made no economic sense.
SAIC was contracted to do the basic design and preliminary feasibility analysis. (Why we didn't do it ourselves given their demonstrated incompetence, is beyond me...) I still remember how in no fewer than probably a dozen places in the document and presentation, their "brilliant" young engineer (who was actually a couple of years older than I was at the time) kept pointing out how absolutely vital it was that SAIC's design used six bolts (all into the same flat plate) to restrain the device in six degrees of freedom. He really believed you needed one bolt per degree to hold something to a flat plate! The document had been reviewed by SAIC's "more experienced" hands, but when we complained about getting crap for our money, they pretty much said, "if you don't like our design, do it yourself". We did, and I've never been willing to work with SAIC since then. (Unfortunately, on of the worst PHB's I've ever encountered in the real world forced a decision to keep far too much of the SAIC design he had spent so much of the company's money on, so the project was doomed. I transferred to another group, and never looked back...)
It's seen an unexpectedly wide adoption since 1983? If it takes that long to get unexpected adoption. how long does a slow rollout take?
No, that's 1993, not 1983. It only seems like twenty years.
My own suspicion is that IPv6 will take off just after the 12th of Never... Seriously, unless the ISP's start to support it, and there is some real end-user benefit (mobility, security, etc.), there's just no valid reason to even think about absorbing the substantial pain of an IPv6 transition. (Face it, it's all most IT organizations can do to get things working with v4. Does anyone really think those folks are going to switch to v6 unless they *have* to? Fat chance.)
After reading your list, it seems that perhaps the issues you have are with IP Masquerading, a special form of NAT where the external addresses are non-exclusive.
This is a great point - I often forget that most people, especially the Linux types hanging around here, have absolutely no idea what real NAT looks like.
They think that Linux' brain-dead IP-masquerading is the same thing as real NAT. It's not. It's not even close.
IP-Masq has lived far too long, and it's time to drive a stake through the heart of that vampire sucking the lifeblood out of our networks. Why no one has bothered to make a real NAT part of mainstream Linux distros and we still suffer on with IP-Masq, I really don't know. Ignorance, I guess...
I thought that everyone had either gone forward to style sheets or back to tables.
Actually, frames are more useful today than ever: In today's world where web pages are more like application interfaces than information displays, frames are a great way to avoid reloading things, because they encapsulate the frame content as a separate URL. DOM/CSS/JavaScript/XSLT and the like will make frames more, not less common. Check some of the more davanced uses of these technologies, and you'll find them wrapped up in frames all over the place...
Fortunately, given the very strong presence of prior art in this case (most especially the Netscape 2.0 browser itself) it seems that there's no reason not to use them. And if they are truly patent protected, it seems to me that if they don't restrict their use in the browsers, they can't reasonably expect to restrict them on the server end... SBC is just up to their usual moneygrubbing games. I'm fully in support of patents, but this abuse is enough to put "Call Birch to change phone service" on my to-do list for later this week...
"You're on an island, after a plane crash in which you were the only survivor. One one side of the island is a village full of maneating cannibals who are starving to death. In the middle of the island are wild boars, jungle cats, and all sorts of poisonous creatures. On your side of the island, there's nothing but rocks and the contents of your survival kit: one roll of toilet paper, some twine, a flaregun with three flares, a can of sardines, and a swiss army knife. What do you do first?"
Actually, the right answer to this one quite clear-cut: Respond with a couple of other questions, based on the following "Rule of threes" for survival (many people start by looking for food, which is a good way to die, since it's actually a fairly low priority from a life-preservation point of view):
Since I've got a can of sardines and some possibility of dining on wild boars, jungle cats, poisonous (hopefully just venomous, and not actually poisonous...) creatures, I've actually got quite a while before I have to start seriously worrying about food.
Assuming they haven't left out crucial information like I'm being crushed under the wreckage of the plane and can't breathe, a reasonable first question might be, then: "What's the weather like?" ("tropical paradise" may well make shelter a no-op), followed by, "Where is the nearest fresh water?" (And perhaps, "Do I have a sheet of plastic in my survival kit to build a solar still?") It's likely in this scenario that water is the number one priority.
Finally, it's important to recognize that things aren't *too* bad, if you've still got luxury items like toilet paper...
While not wanting to turn this into a license fre-for-all, I'll just point out that I think the BSD license is generally superior, but would favor an LGPL derivative (sans the socialist Preamble) under one condition: That the author be able to distribute WITHOUT source code for a period of two or three years, after which the source must be made available. This seems, to me, to strike the best possible balance between commercial expoitation (which is realistically required to drive some significant advancements) and the common good. It's important that there be some period of exclusivity if we expect people to make non-trivial inventments in technology development.
(In one sense, this points to the root problem that patents last for an insanely long time in fast moving technology areas. I've proposed making the term of patents inversely proportional to the number of patents *granted* (not applied for) in that category on a sliding scale from two years at the low end to the current maximum for very slow-moving areas. This mechanism has the advantage of being self-regulating, and extending or contracting patent terms as activity heats up in certain areas.)
Mark my words, for the computing paradigms we have today, we've pretty much hit our limits. Don't hold your breath, any real improvements are 20 years out.
:-)
And we might as well just shut down the Patent Office, too, because everything that can be invented has already been done?
Seriously, I think you're *way* off base here. The technologies to build a kick-butt distributed/networked filesystem have existed for years, but there has been no economic incentive to bring them all together. AFS/XFS/Coda/etc. show that there are a lot of interesting avenues for such development, but this is an area where open source may not be able to lead in terms of wide-scale adoption of such systems. On the other hand, if Microsoft, Sun, and say, Red Hat got togther to build an NFSng that all would put in their products, we'd have a *real* revolution.
The barriers to next-gen filesystems are primarily political and economic, not technical.
in a related study, engineering isn't necessarily the best way to be social.
:-) [ducking behind fireproof object...]
that jerk on the tour that told you chicks dig engineers was a lying bastard.
Actually, chicks do dig engineers. But scientists, especially computer scientists, are another matter, entirely...
Leap seconds may make life as a programmer more difficult (right now, I'm working on figuring out how long-term instrumentation systems should handle thier timestamps when leap seconds occur), but they are absolutely necessary:
The basic problem is that two things have to happen:
It's vitally important for navigation and many other purposes that our clocks be corrected as required to reflect the fact that the *real* units of time aren't some stupid atomic counter, but the positions of the heavenly bodies that are used to confirm all references. In essence, the real problem is that the lab rats don't want to recognize that their ridiculous wiggling atom setup is worse than useless in the real world, since it creates all kinds of synchronization problems and in any case, as Einstein said, we really have no more reason to believe the atomic clock is accurate than that the celestial clock is accurate. It's a relativistic problem - while it's probably more likely that the earth is slowing down, there's no way to prove cesium atoms aren't speeding up, oddly, while other research shows that the speed of light may be slowing down...
Long live leap seconds - at least until we get around to redefining time based on celestial mechanics, which is the only thing that makes any sense in the real world, anyway...
The only replacement for AFS that is even close is Microsoft's "Win2k AD'd dfs", and even it is missing a large number of features that AFS has.
;-)
Most notably the ability to build a distributed filesystem that doesn't have a single point of failure, as was required by the MS DFS the last time I checked, since all the name re-mapping happens on a singel server. Ugh - what were they thinking? This has got to be one of the most egregious examples of Microsoft not thinking through the problem before writing the code.
On the other hand, MS DFS was never intended to be a real solution, but just to keep W2K from being disqualified because it didn't have heirarchical mounts. I've seen MS draw it like a gun in several companies where this was an issue - a situation that really sets me up to earn my consulting fees.
Nothing else is - not NFS, not SMB, not whatever, just NCP.
:-)
You will want to look into AFS then before making such a judgement. It may not change your opinion, but for file serving there is very little lacking.
I first used AFS back when it cost a zillion dollars from Transarc and you had to deploy the whole stinking Athena environment to even use it. (This was years before IBM bought Transarc, but the prices are still ridiculous...) I looked at it hard again last year. It's interesting, and certainly improved, but I'm still not sure it's ready for prime time, especially since there are big chunks of functionality and platform support that are simply MIA.
I really don't hink it's possible for any third party network filesystem to have a major impact these days, since unless it's embedded in the major OSes, it's DOA. If Sun and Microsoft could work togehter for once, we might finally get a decent network filesystem for the 21st century, one with things a good distributed lock manager and fine-grained permissions. Oh, well, I can dream, anyway...
They sold almost nothing, so in that sense they failed miserably. What they *did* do before Red Hat did it, was make it possible to say "Linux" in a meeting with a CIO and be taken seriously. They deserve some credit for that, even if reality didn't match the perception...
Isn't that what I said? (Poor grammar and ambiguous modifiers make following the thread quite difficult. In any case, my point was that BSD, MIT, and X licenses are not only open source (or Open Source, if you prefer), but also truly free, in that they are permissable for commercial use. This is the real difference between the BSD and GPL license camps: the GPL folks are primarily concerned with promoting a political agenda, while the BSD folks just want good code to be usable everywhere, recognizing that a rising tide lifts all boats...
Not exactly, it's a dumb terminal. It is basically a full-featured WinCE powered system with the sole purpose of mirroring what is on the servers display.
Think of it as doing a remote X display, if you are familiar with X11. Most of the gripes about it come with the first run of a new technology (from Microsofts point of view.) I would certainly love something like this that operates using X instead.
Actually, it has one major advantage over a Tablet PC - since it's not really one, and doesn't have to conform to Microsoft's ridiculous hardware profile in order to even be eligible to buy the TabletPC version of XP, it has a *real* touchscreen - one you can touch with your finger instead of a special active pen. In my mind, it's this requirement for an active pen digitizer that is the single stupidest thing abou the Tablet PC hardware profile, and may be a large part of the reason why this platform will not do as well as it should. (People want tablets, and the Tablet PC is the best yet, but still not good enough for a lot of folks...)
BTW - you wouldn't really want one of these that does X - that would be much slower than RDP, which is that rarest of creatures - a Microsoft protocol that's best-in-class.
Another thing that Linux folks should remember, but most of them couldn't spell Linux at the time
:)
:)
Insulting your audience on the first line is rarely a good way to get your point across.
It's really not an insult - there's more than a little truth to it - some of the people so ardently reciting the Linux party line here really *couldn't* spell Linux back then. The Linux community is great at creating converts, but very bad at giving them any historical context, or even acknowledging that (gasp) Linux may not be the world's best OS in some respects. I just thought that was kind of amusing...
You say in one bullet that Caldera was the first serious commercial linux, and in the next bullet talk about Red Hat preceding them. By the time I played with Caldera 1.0, Red Hat was pretty well established. Red Hat shipped their first version in '94, Caldera formed in '94 and didn't ship until some time after.
Red Hat was first, IIRC, but I stand by Caldera being the first *serious* commercial Linux. Red Hat was an early integrated distro, but they weren't viewed (and perhaps didn't yet view themselves) as a viable Linux vendor for the enterprise early on. Unfortunately, Caldera never articulated a really viable business model, and it sickens me to see them abandoning that work in favor of SCO.
Oh well, thank Bill Joy (and Theo de Raadt) for BSD!
eDirectory for Linux has been around for a while, I'll give you that, and Novell has a long (and sad) history with *nix. I'm not sure that the history of Novell and *nix is something to be proud of, however. That would be like talking about how qualified Novell is to contribute to the future of word-processors.
No question, there's been some exceptionally bad management and really bone-headed decision-making at Novell in the past decade. They certainly should have done much better given the ooprtunities they've had. (I think largely that's because they've never come to terms with the need to move to a Unix/Linux style OS as the preferred product platfrom, so even though part of the company recognized they should change, another part was too comfortable to allow it.)
Mainly, though, I just wanted to point out that Linux and Novell are more connected than many people may think at first...
Novell played a pivotal role in making Linux an option to be watched and considered. For those of you that don't know, so long ago that it's hard to find documentation of it on the web, Novell's then-CEO, Ray Noorda, had a skunk works project within Novell to build a revolutionary OS built on a strange free operating system called Linux. (IIRC, it was code-named "Corsair")
This fundamentally changed the course of Linux forever:
Anyway, it seemed to be worth pointing out to those that don't know or remember how Novell and Linux have been related for a decade now, and why it's very good news to see the two renewing their relationship. As Rick said to Captain Renault as they walked in to the mist in Casablanca, "Louie, this could be the beginning of a beautiful friendship..."
Most of what Novell does is rather mature on that level. Much more so then Linux, but probably not as much as he thinks. It has great directory, authentication and network file systems. A good AFS, LDAP, Kerberos run Linux domain is perhaps less of a polished product then Novell, but it is not far behind.
Sorry, but Linux isn't even in the same league when it comes to network services. NetWare has its warts, but so far as NOS capabilities they are in a class of one. (Although Banyan was interesting, are they still around?)
The fine granularity of file permissions in NetWare is an absolute dream, and matches and supports real-world needs *far* better than those of Windows, or especially Unix-derived servers. (I've been dealing with the brain-dead Unix file permissions for 18 years now, and the whole system is a major dog's breakfast.)
ACLs have been grafted onto various network filesystems in myriad incompatible, incomprehensible, and unmanageable ways, but that's really no substitute for a just having a reasonable set of permissions capabilities in the first place.
Further, NDS is far and away the best directory service available today - it's really a shame it hasn't taken hold in the Unix/Linux world, as we need it badly if there is ever to be any hope of holding AD at bay. (Those that don't do serious enterprise work fail to comprehend that it's AD that makes it virtually impossible to pull Windows out of an organization - this is the *real* Kool-Aid, and if your organization has drunk a long draft of it, you're poisoned, bucko... Raw LDAP is not really an option in most environments, as the staffing required to manage it that way exceeds the available talented labor pool in most places...)
You're right that all the apps built on these network services have no real equivalent at all in the Unix/Linux world, and only shabby imitators in the Windows world, but even at the server-only level, NetWare in unequalled. I hate the way you administer it, (it's intentionally obtuse to encourage CNE certification), and it has some weaknesses as an application server, but it works and works well.
It's well-engineered, too: As a protocol jock, I say that with real knowledge - compare the rock-solid reliability, wide area bandwidth efficiency, and latency insensitivity, not to mention advanced features and security of NCP to *anything* else, and I think it will come out *way* ahead. I've built worldwide remote site networks that have to have transparent file access back to civilization via a satellite telephone, (the worst latency environment within three planetary diameters) and only NetWare and NCP are capable of operating in such an environment. Nothing else is - not NFS, not SMB, not whatever, just NCP.
Gee, this sounds like a Novell ad - It's not, I haven't even touched the product in two years, but what they do, they do well.
*shrug* I disagree. Any http caching proxy runs into the same thing. It's highly unlikely that either of these folks will run into legal liability.
But there's on big difference: Any traffic crossing *my* cacheing proxy *is* the responsibility of someone on my network. Not so with Freenet.
Isn't the grandparent saying that a licence which forbids you from selling the software is not free? If so, haven't you just confirmed their point?
:-)
Could well be, I'm not sure I saw that one. That's one of the problems with letting us set our own view preferences...
I din't say dying, but it's not well. Clearly, Linux is going to remina in the embedded space for a long time, but it seems much more current effort on the part of the hardware vendors is going to getting things ready for CE rather than Linux.
I couldn't agree more that Lineo and Metrowerks couldn't justify thier price premiums - and that's kinda the point...
Just *try* and find a small lightweight wireless webpad-type device that's not running CE right now. So far, I count only one, and it's an obsolete product that is not very rugged and not being upgraded.
I don't like what I see, but the trend is undeniable.
This is a business admission that the market share for winCE is, in fact, dropping like a stone. The embedded market uses Linux, which allows the tailoring of capabilities, a general understanding and lots of third party documentation ( like o'Rielly).
Sorry, I'm in the middle of an embedded project requirements specification right now, and you're way off base. Take a look at companies that used to have a notable embeded Linux bias, and many are either rushing to CE (Intrinsyc, Tiqit) or have been rendered ineffective in other ways (i.e., Lineo/Embedix absorbed into the Metrowerks amoeba.)
Embedded Linux and/or BSD are attractive, but it's proven to be a very difficult market to make money in. Well over half the several dozen links I collected six months ago on embedded Linux SBCs either belong to companies that are now defunct, or to companies that are deliberately trying to distance themselves from the profitless Linux market. And these are the *hardware* guys - It's even worse amongst the software companies! (Come on, I bet 90+% of the people here would never consider paying for an embedded Linux distro - the perceived value is just not there, except for a very few customers that can't provide enough revenue to ensure survival.
This is a basic economics problem, and one that may not be resolved in favor of open source. (Although I will show my stripes and say that I think BSD is gaining in the embedded space precisely because it permits recovery of investments in ways that aren't do-able under the GPL. Well that and the fact that NetBSD runs on darn near *anything*...)
What MS are saying with this deal is "we are the only ones allowed to licence WinCE. You can add to it, change it, remove stuff, so long as you tell us. BUT a copy of WinCE is still ours to licence. Sounds fair to me.
This is good news, and it is a completely legitimate way to open source something. One of the author's principal rights that many open source licenses neglect is the right of the original creator to continue to direct the development of code, especially to keep it philosophically in line with the original intent.
Although, like all freedoms, it can be abused, this can be a very good thing, too. It is precisely this sort of control by Sun that has kept Java from fragmenting far worse than it has. Ditto for Netscape and Mozilla. (Remember Mozilla is dual-licensed, a concept that has not been proven valid in court, and may not, since it defines inherent contradictions.)
This is really what all squabbles about intellectual property boil down to: how much control does an author or creator have after the fact?
This announcement is very good news, although I fear it means that MS will now roll through the embedded world virtually unopposed. But on the other hand, if no one else is capable of mounting a defense, then perhaps they shouldn't. CE is architecturally ugly, but it works much better than it used to, and one of these days may have enough features of a real OS to be a contender. As I posted here just recently, all the embedded hardware OEMs (even the ones that started out carrying the Linux flag) are rushing to CE to survive, and *that* trend will only grow with this announcement.