it is stupid of you to present your uninformed conjecture about what goes on at MSR as fact. You don't have any idea what people at MSR do, so why did you post a message at all ? There are so many other ways for a slashdotter to express that they dislike microsoft and anything about microsoft, what made you choose this particular subject ? Just curious.
its funny that you dug through my website and found that. its an excerpt from a book i bought called "The secret meaning of birthdays" or some such. I bought it because i picked a few random people that i knew, and the pages about their birthdays generally seemed to describe them pretty well. To this day i haven't decided if theres "something to it", or if the wording is sufficiently general to let anyone read into it whatever they like. (i.e. "oh, i could see that about her, this book is right on the money!")
I dont have friends or foes. (But apparently some people have marked me as their friend/foe). I don't dislike you, i dont _acutally_ think you're stupid, and you're right, it was lame of me to counteract your hipshot with one of my own.
How about this -- next time you have a question about MSR, i'll point you at the information you want. There are plenty of factual reasons to dislike microsoft, so lets keep this a clean fight:)
i would love to give you a long and detailed explanation of where this "argument" falls down, but its easier to just cut to the chase:
you're stupid.
If you want to understand what MS research is, why dont you visit the web sites ? I'll give you a hint. At a research university, there are a bunch of faculty. They cut their time between doing research and teaching. Big name professors are big names because of their research interests, publications, and sometimes industry connections. Not because they're swell teachers.
MS research is similar. Except nobody wastes any time teaching.
It is typically NOT the case that research in MSR is vagely related to something that can be productized. It is PURE computer science research, in a wide variety of areas. Sometimes, some of this research has huge upside for something MS wants to do with a product. Sometimes that takes years. You should browse through the different projects and talks on the MSR site, and ask yourself how much of that you see in MS products today.
As far as "creative freedom", well, i happen to know a few people at MSR that are doing their work on linux, because they feel like it.
there is an awful lot of folklore about what is and isn't windows NT 3.x, 4.0, w2k, and xp. one thing slashdotters are notoriously bad about is taking 3rd hand information that they _want_ to beleive and retelling it as gospel.
this gets extra bad when a moderately well-informed poster gets a few things right but then goes haywire elsewhere in the post.
Windows NT at its core is not VMS, not Windows, Not OS/2, and not anything else. It is its own thing. It is(was) highly portable, built with many modernish OS concepts, designed for SMP/fine grained security from the beginning.
This is the core of NT. You wouldn't have any fun writing an "NT program". But some do exist - for example, autochk.exe (the w2k chkdsk that runs in vga-console mode on bootup) is an NT "Native" application, because it runs too early in the bootup to use any of the subsystems.
Which is a good lead into what most people know of NT. At the time NT was getting started, IBM had this OS/2 1.3 investment. IBM wanted to move forward with OS/2, but was in a deal with Microsoft to make it happen. So one of the goals of NT was that it would host OS/2 1.x applications. Thus, the OS/2 subsystem. IBM gets a new OS from Microsoft that their existing customer base can use their existing apps with.
Back then, you weren't an OS unless you were POSIX compliant. More importantly, you weren't eligible to win certain government bids if you weren't POSIX. So NT needed to have a POSIX interface to it. Enter the POSIX subsystem. With a POSIX layer ontop of the NT system, MS can say"look at this POSIX OS we've got!"
Eventually, IBM and MS's dealings got sour. MS say that Win16 was really doing quite well, and Win32 was actually materializing. And OS/2 1.x wasn't going anywhere. Suddenly, it becomes important to flesh out another implementation of Win32, and bring that forward into NT instead of relying on OS/2 programs. Enter the Win32 subsystem.
So we've got NT, which is hosting 3 subsystems - OS/2, POSIX, and Win32.
These subsystems are completely insulated operating environments. They use the NT provided interfaces to interact with the operating system - which is NT. Programs are written against the NT apis, but more usually, the subsystem APIs. The subsystems expose the NT functionality (and extend it) in different ways, as required.
If you find an NT 3.x box, that was back when the subsystems were all sort of "equal parties" -- all clients of the NT operating system. As time went on, people figured out that there was no OS/2 software that mattered, and getting better performance out of Win32 on NT was going to be important if anyone was ever going to use NT. So for NT4, more of Win32 went kernel mode... i.e., the Win32 subsystem got some help. A lot of people bitched about this, saying it would lead to decreased stability, etc etc. I even bitched about this in my MS interviews. Turns out i was talking shit (like most people do when they read a lot about NT in trade rags and websites) One of the early chapters of "Inside Windows 2000" addresses this point exactly - moving more of Win32 into kernel mode didn't realy matter, because if the win32 subsystem died, the box would force-panic anyway.
Anyhow, these days in windows xp, the posix subsystem is all but gone, the os/2 subsystem is gone, and the win32 subsystem has been heavily "favored" for performance. But its still largely a client of the NT API.
One caveat is that if you get "Services For UNIX", the POSIX subsystem gets replaced with a fully featured unix/posix environment. You get a real bash/tcsh, making posix calls to something that looks like a unix kernel - but its really just a fully fleshed out subsystem making calls to the NT APIs. SFU is one of the best things any person who likes the commandline power and tools of unix can do for themselves if they use a W2k or XP box. I regularly pipe data between excel and one-off awk/sed/perl/sort/uniq constructs that do just what i need them to.
One of the things thats surprised me the most about NT is just how much is going on under the covers. You've got the entire complexity of the Win32 api (that most programs are written to), which in turn makes calls to the NT api, which at its core is stilly very rich and passes around lots of objects of different types.
Incidentally, you can see this subsystem stuff on any W2k or XP box. Open up task manager, and find smss.exe. Thats the session manager, which starts the approprate subsystem (if required) based on a flag (i beleive) inside the header of an executable. You'll definitely have csrss.exe running, which is client/server runtime subsystem. Thats Win32. It is a child of smss.exe, and if csrss.exe ever returns, smss bluescreens the box.
If you install SFU and run some SFU programs, there will be a POSIX.EXE and a PSXRUN.EXE. These are peices of the SFU posix subsystem layer. ANd if you run suitably old win16 or win32 apps (something like the VB6 installer) you'll see WOWEXEC.EXE show up, with an ntvdm.exe i beleive. YOu'll see this immediately because the name is indented in task manager and doesn't have any data displayed in the process list.
An interesting point is that NT has had a lot of what unix people say windows lacks featurewise for a long time. At its core, NT is in basically all ways more advanced than UNIX. The problem comes in the layering and the adoption. Your only interface to NT is Win32. Win32 has existed in around 5 implementations, afaik (hosted on windows 3.x, Win32s, Win95, NT4, etc). They actually have to be "compatible" with each other. NT provides so much more featurewise (and perhaps pitfall wise) than the old Win16 environment did (which Win32 evolved from).
So given that at the Win32 layer, you're already basically filtering out the functionality that Win32 doesn't conveniently expose, now you must deal with applications that were written in the "old mindset". Single user, full priviledges, non-existant memory protection. If some new feature gets taken in the NT core apis, it must then be exposed in the subsystem, then in the win32 apis, and then finally in win32 applications. That takes time - even within MS.
Despite all of this, NT has demonstrated surprising adaptability. It's roughly 10 years old, and its binary compatible with applications that pre-date it by 15 years. Despite all of the layering and complexity, it performs pretty well (look at tpc and tpc/c benchmarks, for instance).
Comparatively, UNIX has been around for over 25 years, and while its certainly bigger and badder than it used to be, in many ways it shows its ancestry. First and foremost is the security model. Yeah, people knock MS for security issues. UNIX has had 25+ years to get the implementation of a simple design right (and it still isn't). The difficulty of writing suid 0 software in unix is well known, and this problem is inherent in the design. Another big deal is the strong marraige of ANSI, C, and UNIX. The NT interfaces are all 16bit char natively, so supporting UNICODE UCS-2 comes for free practically. For W9x ANSI compat reasons, Win32 duplicates most of the interfaces with ANSI (8bit char) equivalents. So much of UNIX, the c library, and the userland are deeply rooted in the idea of char being 8 bits that unicode will probably never happen in a decent way (UTF-8 is nightmare, its up to each app to do the right thing, etc).
Ok, i suspect this has hit the comment length filter. If you're interested in the guts of NT at all, the w2k and xp debugging symbols and kernel debugger are freely downloadable. Turns out its quite easy to get quite a lot of info from a closed source OS if you've got the debugger. Better yet, if you get "Inside Windows 2000" you'll get far more info than you could dream of on how NT works, and it includes quite a few good tools for really understanding whats going on underneath the clicks and buttons.
screw the environment, screw the epa, and screw people that beleive this bullshit wholeheartedly.
If i got an electric vehicle, should i be permitted to go 90mph ?
The NHSTA (or whatever) is a bunch of tools, is the big deal.
It is a known fact that fuel spent is roughly analgous to horsepower produced, and that the power required to maintain a constant speed in the face of air resistance increases as the square of the velocity. Thus, it takes more gas to cruise at 65 than at 55, and the difference between 65 and 55 is more than the difference between 45 and 55.
That said, im already taxed on gas. Quite a lot.
If the government cared a lick about emissions from vehicles, SUV's would be outlawed, and trucks would be taxed silly, and those taxes would be inconsequential because trucks are a cost of doing business. People own trucks for three reasons 1) they need the utility of a truck for their profession 2) they need utility of a truck for their leisure 3) someone urinated in their mother, with the resultant genetic effects
highly efficient european sports cars pay ridiculous gas guzzler taxes because they happen to have some power in them. Yet SUV's and trucks do not pay these same taxes, regardless of their use.
The #1 thing the US government _should_ be doing with highway safety is getting rid of older cars and SUVs, as both have terrible emissions, terrible brakes, and terrible overall safety.
The speed limit is a joke. It's what happens when a bunch of "Kyle's Moms" (from southpark) get to gether and make laws.
Look at the passenger miles driven and fatalities in the USA and germany. They're roughly the same. Yet germany has no posted limit on 30% of its motorways.
(and significantly fewer vehicular deaths resulting from alcohol, despite being the home of thousands of distinct beers)
Point 1) Dropping a computer is a dumb idea regardless of whats in it
Point 2) All musicians of any caliber with amplified string instruments use Tube amplifiers which are driven on shitty trucks in steel road cases thousands of miles every year. They dont remove the tubes before they box em up. The tubes in the amps are _quite_ difficult to get in and out of the sockets, and they hold up remarkably well.
The tubes on a motherboard are probably the least of your durability concerns, motion-shock damage wise.
You're right, a solid state device of sufficient clocking and fidelity could provide a 100% accurate wave recreation, with zero distortion.
As far as what _audiophiles_ want, i dont give a shit - they're all insane and have too much money.
Where people _really_ like tubes is in guitar and bass amplication. And the reason is that tubes in overdriven conditions sound "better" than solid state "distortion" (which is really just clamping, fundamentally)
The technical reason most often cited for tubes sounding warmer has to do with the order of harmonics that tube amplification creates vs solid state. I beleive tubes make 3rd and 5th order harmonics that sound more "ear pleasing".
IMO, tubes for computer audio on a motherboard is one of the dumbest ideas ever, technically, but one of the best ideas ever, marketing wise.
The microsoft balance sheet is a matter of public record. I asked you where your data was because i know you're making it up.
Your statement that MS spends more on marketing than _anything_ else is ridiculous. Paying those employees has to cost something, dont you think ?
Here's an example. The entire world made a huge deal when they found out that the XBox marketing budget was going to be 500m dollars. 500m is a drop in the bucket, budget wise. And it's been one of the most aggressive advertising campaigns ever.
Your bit about including items in other budget "buckets" that are really marketing type moves (i.e. writing off stuff you give away) is moderately insightful, but comes out to a rounding error in the overall microsoft budget.
Nothing about what you said is "obvious". And given how much money we throw at lab hardware on my team, (and how little we spend on marketing), im pretty inclined to say you're full of it.
Your entire original post is 100% presumption. And your follow up is worse. I don't have (or want) an MBA, nor do i care about marketing. If i found out that MS spent more on marketing than anything else, I'd be more upset about it than you'd be, beleive me.
In any case, i can only conclude that you want to be known for your baseless statements which evolve into presumptuous insults, as you've presented nothing else in this discussion.
Here's some data for you, though: http://www.microsoft.com/msft/sec/FY02/10 k2002.htm
or, to spell it right out:
Sales and Marketing. Sales and marketing expense as a percentage of revenue was 18.0% in 2000, 19.3% in 2001, and 19.1% in 2002.
Thats _all sales and marketing expenses_ 19.1% of 28b is a hair under $5b.
I happen to know that MS spends significantly more than $5b on R&D, for example. Much more so than any "free stuff writeoffs" could possibly make up for.
Actually, it does't require years of training, hard work, vitamins, or anything else.
See, thanks to EP1, it turns out that the force was all along a system of genetic advantage backed class superiority. Gone is the happy message of "work hard, do the right thing, and you will be honored".
Now, it's "hope you had the right parents, dirtbag"
Ok. Here's a concrete place that IIS spanks unix+apache.
You have 100 vdomains. Each of them gets to run {cgi|perl|php}.
You want them to run as separate user credentials, so nobody pees in anybody elses pond.
Whats your answer ?
Hint: It's su-exec. Which requires you compiled php as standalone. Which negates any perf benefits of php. Which has its own security consequences and other "gotchas".
On IIS ? Right click "app properties", change the security credential used for "anonymous access".
(You can also script against the IIS metabase, iirc, if you dont want to use the GUI method)
If you're writing an app that uses ASP.NET, its even easier, just specify the impersonation crednetials in the Web.Config file of the aspx project.
actually 32 bit apps on SGI's (and other systems) will run faster than 64bit software, because the size of a pointer is half as large, so you can fit more pointers in a dcache line, for instance, and there are no instructinos taking 64 bit memory addresses (which are naturally twice as large) (although very often do you use one of the ld/st instructinos that specifies the full address)
now, SGI complicates this a bit, theres o32, n32, and 64 bit binary ABIs. Most everything is actually n32 - 32 bit address space, 32 bit instructions, but using the MIPS4 ISA (r5k, r10k and later)
"if you buy an animal from a store, you are not allowed to modify it"
but then if someone else makes a law saying
"if you buy a dvd player from the store you are not allowed to modify it".. then people get all pissed off about it ? THe usual argument here is "its mine, i can do whatever i want with it"
why does that not apply to animals ?
(note that i dont advocate torturing animals, beating wives, or stealing software)
on the other hand, if i got pissed off and kicked my dog, it seems ridiculous that i'd probably get a stiffer sentence than if i date-raped a girl (especially if i was in the same frat as the prosecuting lawyer was when he was a kid)
fwiw, i agree with you. going to jail over this seems silly when people that commit violent crimes are getting let out of jail to make room. but there should be _some_ penalty. one that makes it sufficiently unattractive that people stop doing it so flagrantly.
Yeah. It was a god-spoof of the looney toons things where anvils would just randomly drop on people.
God's mighty anvil was great.
Something stupid/naughty would be going on "down on earth". Then you'd see a scene change to god walking around in a cloud, looking down, holding an anvil. he's squint downwards and line up the anvil just right, then drop it.
Then it would land on whoever was down on earth causing the issue of interest.
if you succeed in killing smss.exe, the machine goes away:)
similarly, if csrss.exe exits, smss.exe bluescreens the machine.
lsass is the local security agent subsystem server. (i always read this is "ls ass"
SMSS is the session management subsystem. it spawns Csrss.exe (Client Server Run Time SubSystem - the Win32 layer on top of NT)
If you have a suitably old smss.exe, it also spawns the OS/2 1.x layer or the POSIX layer. If you have Services for UNIX, there is a new posix.exe layer and psxrun.exe servers that you'll also see.
No you dont. Irix 6.3 was For O2's only. You're either running 5.x, 6.2, or 6.5 on your Indigo2. If its got the R8000 (power Indigo2) it might be running 6.01 or 6.1
Or, you have an O2, not an indigo2, in which case you're running 6.3 or 6.5
Well, i dont blame them. The 100mbit cards for the machines that didn't ship with 10/100 were expensive and SHITTY. For instance, for the indigo2, you could get an EISA phobos ethernet card which is basically a 3c597 with a different rom. This was spendy and you would never saturate the 100mbit card and it would generate a crapload of interrupts (slam cpu utilization).
Similar story for SGI Indy/Challenge S. The 100mbit workstation cards were 3rd party, spendy, and not so hot.
you could get SGI labeled 100mbit cards for the VME boxes (Challenge DM, for isntance) that worked pretty well.
well, noone will _make_ me apply that policy to my emails or images. Now i cant choose to do so if i want to. Tomorrow I may be able to. Palladium wouldn't restrict my ability to send email. It doesn't "restrict" anything - It lets ME restrict your ability to do things i dont want you doing with my stuff, which is _Exactly_ what I do want.
The point is that the control of the content usage should be at the hands of its creators. The _only_ caveat is preserving fair use for copyrighted works. One advantage of big media companies being big media companies is if they categorically inhibit fair use by making bad DRM policies, they'll get spanked for it legally, eventually. The recent price fixing suit should be evidence enough that they're not the golden children of capitol hill.
I don't want you to have fair use of my emails, my photo album, or anything else thats mine. I want you to have exactly the usage that I say you have.
Maybe with palladium in place, the DMCA can be overturned ? Palladium doesn't attempt to restrict your ability to create software (DMCA does). It makes it technically hard (hopefully infeasable) to circumvent restrictions placed on digital information. Hopefully, legality will be removed from the picture altogether, which means dumb lawmakers wont be making dumb laws.
THat means that the dmca can go, and you'll be allowed to write your software dvd player. ANd if future dvds are released for palladium-only devices, too bad. Hollywood isn't obligated to release a DVD at all. When you're making better movies (which really wouldn't be much of a stretch, honestly;) you're certainly free to release them in whatever unencumbered format you like. Palladium wont prevent that. It might even be beneficial to you.
You're certainly free to develop a non-tcpa and non-palladium alternative for every widget in the stack, from cpu to web brwoser and everything inbetween. (unless fritz gets his way). you can't realistically get upset if nobody wants to use them besides you and a few slashdotters.
In a way, it's similar to the "this page doesn't work with opera on linux" argument. Sure, web designers probably _should_ test sites on every browser ever. But they dont. You have no right to tell them how to make their stupid site. You can just choose not to go there.
If you've tried visiting a website using Omniweb 2.x on NextStep, you know how useless much of the web is unless you're running a recent browser on a supported platform.
Ideological living is expensive and inconvenient. If palladium/tcpa have merits in the eyes of adopters(and it looks like thats a possibility), how can you plausibly be upset about its adoption ? Because it might inconvenience you ?
Palladium isn't about jpegs, picts, movies, etc... it's about controlling the types of software that can run on your computer.
Please explain why you think that is the case.
you don't want everyone seeing your picture, encrypt it and send it to a friend.
If i encrypt a picture and send it to a friend, they simply unencrypt it and send it to some other "friend".
Without a comprehensive DRM scheme, there is no way to "give with restrictions". It's either give, or dont give.
Point taken about outputting the image to a DVI panel. You do know that there are provisions for anti-sniff DVI work, right ?
MS doesn't want to make an MS computer, because there's no money in hardware, and they know it. There is money in software. The xbox is a platform that puts MS in a good position to make lots of money on software - games.
I've had conversations with a few of the people actually working at MS on the drm/palladium stuff. It's funny that you're trying to tell me what MS's plans are in this regard.
I for one am really looking forward to having a palladium infrastructure in place.
1) I wont buy shitty music regardless of how its encumbered, so that doesn't effect me.
2) DVDs are encumbered now, but i can do whatever i want to with them (watch them). There are black-market players that are available to me that let me skip commercials, undo region coding, etc, but as much as i dislike those attributes, i haven't voted with my dollars (yet)
3) If someone like microsoft (or apple, or sun, or any other technology company) outlines how DRM et al should work, that means record companies and legislators _ARENT_ doing it. When was the last time you ran across a good government standard ? When was the last time you saw a peice of software by a media company that was worth using ?
4) I really like the idea of being able to put a usage policy on content. For instance, if i scan some pictures of a vacation, and put them on a website, perhaps i want to apply a policy that says these can only be viewed by people that have access the picture from the URL. Then i can simply restrict access to the site via the normal means, and the pictures will not be redistributable once they've been downloaded by the viewers that I do allow.
you think this is stupid, and maybe it is. On the other hand, a picture taken of my wife and a friend of ours (totally clean, of course) from a party we went to ended up on a "hot teen of the day" site. Asking the site adminstrator to take the pciture down of course was a pain. And i was lucky that they complied.
If there had been some sort of DRM policy on JPEGs then i wouldn't have to be upset with the friend that posted the pictures on his site for not restricting access. I wouldn't have to flinch everytime i see a camera to think "where the fuck is this going to end up, maybe with a few photoshop edits"
The MS work behind TCPA/Palladium is not as orweillian as you think. It is not a framework designed to let hollywood hold all the strings (nor is it setup to let MS hold all the strings).
TCPA / Palladium do not add any restrictions. They allow new types of things to be done that cannot currently be done (sending an email that cannot be forwarded, distributing an image that cannot be edited, etc)
If media companies stop distributing media in clear-text, thats their choice. Vote with your dollars. If, despite rational arguments to the contrary, you cannot stand the idea of TCPA, you'll be able to vote with your dollars there, as well.
The throttle in the E39 M5 and E46 M3 is electronic. They are drive-by-wire. There is no direct mechanical linkages.
Not that that matters. in my 88 BMW, i _do_ have a direct throttle cable (i've even replaced it), but i still don't need to retard timing to limit acceleration. The hard rev limiter is a fuel cut off. No timing modifications necessary. The momentary rev limit is a temporary fuel cut.
Naturally fuel cutoffs are probably too jerky for traction control applications, but they're definitely used for rev limiting. Most TC systems that need to actuate a slow-down condition use the brakes more than ECU work, however.
Actually i think the spark-retard thing has to do more with air temperature. If you read what Dinan has to say about their tuning on the E39 M5 package, they found that alot of the power loss comes from intake air temperature. The the ECU takes intake air temp as one of its inputs, and it knows that as the temperature of the air rises, it cant be as aggressive with its timing, so it slwos down. The biggest changes dinan made to the E39 M5 were on the intake, to keep the air as cool as possible. This lets the factory programming make the most of things while still staying inside the safety envelope.
Clearly running a car on the dyno thats _expecting_ 130mph of cool air coming at its radiator and intake is bad, and artificial air is a serious requirement.
I'm surprised that you say that the ECU retards the timing based on wheel speed. Many BMWs of course have traction control (and even the 80s bmws had 4 channel ABS) and clearly that cares about wheel speed differences. The TCU needs to talk to the ECU to tell it to cut throttle when wheel(s) spin, but it wouldn't do that with spark retard. Do you have any pointers to a discussion of the front-wheels-stationary problem ?
Slashdot Mirror
it is stupid of you to present your uninformed conjecture about what goes on at MSR as fact. You don't have any idea what people at MSR do, so why did you post a message at all ? There are so many other ways for a slashdotter to express that they dislike microsoft and anything about microsoft, what made you choose this particular subject ? Just curious.
:)
its funny that you dug through my website and found that. its an excerpt from a book i bought called "The secret meaning of birthdays" or some such. I bought it because i picked a few random people that i knew, and the pages about their birthdays generally seemed to describe them pretty well. To this day i haven't decided if theres "something to it", or if the wording is sufficiently general to let anyone read into it whatever they like. (i.e. "oh, i could see that about her, this book is right on the money!")
I dont have friends or foes. (But apparently some people have marked me as their friend/foe). I don't dislike you, i dont _acutally_ think you're stupid, and you're right, it was lame of me to counteract your hipshot with one of my own.
How about this -- next time you have a question about MSR, i'll point you at the information you want. There are plenty of factual reasons to dislike microsoft, so lets keep this a clean fight
i would love to give you a long and detailed explanation of where this "argument" falls down, but its easier to just cut to the chase:
you're stupid.
If you want to understand what MS research is, why dont you visit the web sites ? I'll give you a hint. At a research university, there are a bunch of faculty. They cut their time between doing research and teaching. Big name professors are big names because of their research interests, publications, and sometimes industry connections. Not because they're swell teachers.
MS research is similar. Except nobody wastes any time teaching.
It is typically NOT the case that research in MSR is vagely related to something that can be productized. It is PURE computer science research, in a wide variety of areas. Sometimes, some of this research has huge upside for something MS wants to do with a product. Sometimes that takes years. You should browse through the different projects and talks on the MSR site, and ask yourself how much of that you see in MS products today.
As far as "creative freedom", well, i happen to know a few people at MSR that are doing their work on linux, because they feel like it.
there is an awful lot of folklore about what is and isn't windows NT 3.x, 4.0, w2k, and xp. one thing slashdotters are notoriously bad about is taking 3rd hand information that they _want_ to beleive and retelling it as gospel.
this gets extra bad when a moderately well-informed poster gets a few things right but then goes haywire elsewhere in the post.
Windows NT at its core is not VMS, not Windows, Not OS/2, and not anything else. It is its own thing. It is(was) highly portable, built with many modernish OS concepts, designed for SMP/fine grained security from the beginning.
This is the core of NT. You wouldn't have any fun writing an "NT program". But some do exist - for example, autochk.exe (the w2k chkdsk that runs in vga-console mode on bootup) is an NT "Native" application, because it runs too early in the bootup to use any of the subsystems.
Which is a good lead into what most people know of NT. At the time NT was getting started, IBM had this OS/2 1.3 investment. IBM wanted to move forward with OS/2, but was in a deal with Microsoft to make it happen. So one of the goals of NT was that it would host OS/2 1.x applications. Thus, the OS/2 subsystem. IBM gets a new OS from Microsoft that their existing customer base can use their existing apps with.
Back then, you weren't an OS unless you were POSIX compliant. More importantly, you weren't eligible to win certain government bids if you weren't POSIX. So NT needed to have a POSIX interface to it. Enter the POSIX subsystem. With a POSIX layer ontop of the NT system, MS can say"look at this POSIX OS we've got!"
Eventually, IBM and MS's dealings got sour. MS say that Win16 was really doing quite well, and Win32 was actually materializing. And OS/2 1.x wasn't going anywhere. Suddenly, it becomes important to flesh out another implementation of Win32, and bring that forward into NT instead of relying on OS/2 programs. Enter the Win32 subsystem.
So we've got NT, which is hosting 3 subsystems - OS/2, POSIX, and Win32.
These subsystems are completely insulated operating environments. They use the NT provided interfaces to interact with the operating system - which is NT. Programs are written against the NT apis, but more usually, the subsystem APIs. The subsystems expose the NT functionality (and extend it) in different ways, as required.
If you find an NT 3.x box, that was back when the subsystems were all sort of "equal parties" -- all clients of the NT operating system. As time went on, people figured out that there was no OS/2 software that mattered, and getting better performance out of Win32 on NT was going to be important if anyone was ever going to use NT. So for NT4, more of Win32 went kernel mode... i.e., the Win32 subsystem got some help. A lot of people bitched about this, saying it would lead to decreased stability, etc etc. I even bitched about this in my MS interviews. Turns out i was talking shit (like most people do when they read a lot about NT in trade rags and websites) One of the early chapters of "Inside Windows 2000" addresses this point exactly - moving more of Win32 into kernel mode didn't realy matter, because if the win32 subsystem died, the box would force-panic anyway.
Anyhow, these days in windows xp, the posix subsystem is all but gone, the os/2 subsystem is gone, and the win32 subsystem has been heavily "favored" for performance. But its still largely a client of the NT API.
One caveat is that if you get "Services For UNIX", the POSIX subsystem gets replaced with a fully featured unix/posix environment. You get a real bash/tcsh, making posix calls to something that looks like a unix kernel - but its really just a fully fleshed out subsystem making calls to the NT APIs. SFU is one of the best things any person who likes the commandline power and tools of unix can do for themselves if they use a W2k or XP box. I regularly pipe data between excel and one-off awk/sed/perl/sort/uniq constructs that do just what i need them to.
One of the things thats surprised me the most about NT is just how much is going on under the covers. You've got the entire complexity of the Win32 api (that most programs are written to), which in turn makes calls to the NT api, which at its core is stilly very rich and passes around lots of objects of different types.
Incidentally, you can see this subsystem stuff on any W2k or XP box. Open up task manager, and find smss.exe. Thats the session manager, which starts the approprate subsystem (if required) based on a flag (i beleive) inside the header of an executable. You'll definitely have csrss.exe running, which is client/server runtime subsystem. Thats Win32. It is a child of smss.exe, and if csrss.exe ever returns, smss bluescreens the box.
If you install SFU and run some SFU programs, there will be a POSIX.EXE and a PSXRUN.EXE. These are peices of the SFU posix subsystem layer. ANd if you run suitably old win16 or win32 apps (something like the VB6 installer) you'll see WOWEXEC.EXE show up, with an ntvdm.exe i beleive. YOu'll see this immediately because the name is indented in task manager and doesn't have any data displayed in the process list.
An interesting point is that NT has had a lot of what unix people say windows lacks featurewise for a long time. At its core, NT is in basically all ways more advanced than UNIX. The problem comes in the layering and the adoption. Your only interface to NT is Win32. Win32 has existed in around 5 implementations, afaik (hosted on windows 3.x, Win32s, Win95, NT4, etc). They actually have to be "compatible" with each other. NT provides so much more featurewise (and perhaps pitfall wise) than the old Win16 environment did (which Win32 evolved from).
So given that at the Win32 layer, you're already basically filtering out the functionality that Win32 doesn't conveniently expose, now you must deal with applications that were written in the "old mindset". Single user, full priviledges, non-existant memory protection. If some new feature gets taken in the NT core apis, it must then be exposed in the subsystem, then in the win32 apis, and then finally in win32 applications. That takes time - even within MS.
Despite all of this, NT has demonstrated surprising adaptability. It's roughly 10 years old, and its binary compatible with applications that pre-date it by 15 years. Despite all of the layering and complexity, it performs pretty well (look at tpc and tpc/c benchmarks, for instance).
Comparatively, UNIX has been around for over 25 years, and while its certainly bigger and badder than it used to be, in many ways it shows its ancestry. First and foremost is the security model. Yeah, people knock MS for security issues. UNIX has had 25+ years to get the implementation of a simple design right (and it still isn't). The difficulty of writing suid 0 software in unix is well known, and this problem is inherent in the design. Another big deal is the strong marraige of ANSI, C, and UNIX. The NT interfaces are all 16bit char natively, so supporting UNICODE UCS-2 comes for free practically. For W9x ANSI compat reasons, Win32 duplicates most of the interfaces with ANSI (8bit char) equivalents. So much of UNIX, the c library, and the userland are deeply rooted in the idea of char being 8 bits that unicode will probably never happen in a decent way (UTF-8 is nightmare, its up to each app to do the right thing, etc).
Ok, i suspect this has hit the comment length filter. If you're interested in the guts of NT at all, the w2k and xp debugging symbols and kernel debugger are freely downloadable. Turns out its quite easy to get quite a lot of info from a closed source OS if you've got the debugger. Better yet, if you get "Inside Windows 2000" you'll get far more info than you could dream of on how NT works, and it includes quite a few good tools for really understanding whats going on underneath the clicks and buttons.
Sorry. I watched Amilee last weekend and the mean grocer uses that line in the movie. It was the funniest thing I had ever heard.
screw the environment, screw the epa, and screw people that beleive this bullshit wholeheartedly.
If i got an electric vehicle, should i be permitted to go 90mph ?
The NHSTA (or whatever) is a bunch of tools, is the big deal.
It is a known fact that fuel spent is roughly analgous to horsepower produced, and that the power required to maintain a constant speed in the face of air resistance increases as the square of the velocity. Thus, it takes more gas to cruise at 65 than at 55, and the difference between 65 and 55 is more than the difference between 45 and 55.
That said, im already taxed on gas. Quite a lot.
If the government cared a lick about emissions from vehicles, SUV's would be outlawed, and trucks would be taxed silly, and those taxes would be inconsequential because trucks are a cost of doing business. People own trucks for three reasons
1) they need the utility of a truck for their profession
2) they need utility of a truck for their leisure
3) someone urinated in their mother, with the resultant genetic effects
highly efficient european sports cars pay ridiculous gas guzzler taxes because they happen to have some power in them. Yet SUV's and trucks do not pay these same taxes, regardless of their use.
The #1 thing the US government _should_ be doing with highway safety is getting rid of older cars and SUVs, as both have terrible emissions, terrible brakes, and terrible overall safety.
The speed limit is a joke. It's what happens when a bunch of "Kyle's Moms" (from southpark) get to gether and make laws.
Look at the passenger miles driven and fatalities in the USA and germany. They're roughly the same. Yet germany has no posted limit on 30% of its motorways.
(and significantly fewer vehicular deaths resulting from alcohol, despite being the home of thousands of distinct beers)
Point 1) Dropping a computer is a dumb idea regardless of whats in it
Point 2) All musicians of any caliber with amplified string instruments use Tube amplifiers which are driven on shitty trucks in steel road cases thousands of miles every year. They dont remove the tubes before they box em up. The tubes in the amps are _quite_ difficult to get in and out of the sockets, and they hold up remarkably well.
The tubes on a motherboard are probably the least of your durability concerns, motion-shock damage wise.
It's not a theory.
You're right, a solid state device of sufficient clocking and fidelity could provide a 100% accurate wave recreation, with zero distortion.
As far as what _audiophiles_ want, i dont give a shit - they're all insane and have too much money.
Where people _really_ like tubes is in guitar and bass amplication. And the reason is that tubes in overdriven conditions sound "better" than solid state "distortion" (which is really just clamping, fundamentally)
The technical reason most often cited for tubes sounding warmer has to do with the order of harmonics that tube amplification creates vs solid state. I beleive tubes make 3rd and 5th order harmonics that sound more "ear pleasing".
IMO, tubes for computer audio on a motherboard is one of the dumbest ideas ever, technically, but one of the best ideas ever, marketing wise.
The microsoft balance sheet is a matter of public record. I asked you where your data was because i know you're making it up.
0 k2002.htm
Your statement that MS spends more on marketing than _anything_ else is ridiculous. Paying those employees has to cost something, dont you think ?
Here's an example. The entire world made a huge deal when they found out that the XBox marketing budget was going to be 500m dollars. 500m is a drop in the bucket, budget wise. And it's been one of the most aggressive advertising campaigns ever.
Your bit about including items in other budget "buckets" that are really marketing type moves (i.e. writing off stuff you give away) is moderately insightful, but comes out to a rounding error in the overall microsoft budget.
Nothing about what you said is "obvious". And given how much money we throw at lab hardware on my team, (and how little we spend on marketing), im pretty inclined to say you're full of it.
Your entire original post is 100% presumption. And your follow up is worse. I don't have (or want) an MBA, nor do i care about marketing. If i found out that MS spent more on marketing than anything else, I'd be more upset about it than you'd be, beleive me.
In any case, i can only conclude that you want to be known for your baseless statements which evolve into presumptuous insults, as you've presented nothing else in this discussion.
Here's some data for you, though:
http://www.microsoft.com/msft/sec/FY02/1
or, to spell it right out:
Sales and Marketing. Sales and marketing expense as a percentage of revenue was 18.0% in 2000, 19.3% in 2001, and 19.1% in 2002.
Thats _all sales and marketing expenses_
19.1% of 28b is a hair under $5b.
I happen to know that MS spends significantly more than $5b on R&D, for example. Much more so than any "free stuff writeoffs" could possibly make up for.
Microsoft spends big when it comes to it's PR and marekting firms. More than on any other single item in the budget.
care to substantiate this ? it would be news to me.
dude, the whole point of redheads is "no ruphies required".
:)
what the hell is wrong with _you_ ? do your teeth look like the back of a hammer or something ?
uh, because they all run with the same unix security credentials. (because they all run with the prives of the apache server process)
Actually, it does't require years of training, hard work, vitamins, or anything else.
See, thanks to EP1, it turns out that the force was all along a system of genetic advantage backed class superiority. Gone is the happy message of "work hard, do the right thing, and you will be honored".
Now, it's "hope you had the right parents, dirtbag"
Ok. Here's a concrete place that IIS spanks unix+apache.
You have 100 vdomains. Each of them gets to run {cgi|perl|php}.
You want them to run as separate user credentials, so nobody pees in anybody elses pond.
Whats your answer ?
Hint: It's su-exec. Which requires you compiled php as standalone. Which negates any perf benefits of php. Which has its own security consequences and other "gotchas".
On IIS ? Right click "app properties", change the security credential used for "anonymous access".
(You can also script against the IIS metabase, iirc, if you dont want to use the GUI method)
If you're writing an app that uses ASP.NET, its even easier, just specify the impersonation crednetials in the Web.Config file of the aspx project.
actually 32 bit apps on SGI's (and other systems) will run faster than 64bit software, because the size of a pointer is half as large, so you can fit more pointers in a dcache line, for instance, and there are no instructinos taking 64 bit memory addresses (which are naturally twice as large) (although very often do you use one of the ld/st instructinos that specifies the full address)
now, SGI complicates this a bit, theres o32, n32, and 64 bit binary ABIs. Most everything is actually n32 - 32 bit address space, 32 bit instructions, but using the MIPS4 ISA (r5k, r10k and later)
Is this an article from The Onion ?
Come on. What kind of headline is that ?
why is torturing animals bad ?
.. then people get all pissed off about it ?
why is it reasonable to make a law that says
"if you buy an animal from a store, you are not allowed to modify it"
but then if someone else makes a law saying
"if you buy a dvd player from the store you are not allowed to modify it"
THe usual argument here is "its mine, i can do whatever i want with it"
why does that not apply to animals ?
(note that i dont advocate torturing animals, beating wives, or stealing software)
on the other hand, if i got pissed off and kicked my dog, it seems ridiculous that i'd probably get a stiffer sentence than if i date-raped a girl (especially if i was in the same frat as the prosecuting lawyer was when he was a kid)
fwiw, i agree with you. going to jail over this seems silly when people that commit violent crimes are getting let out of jail to make room. but there should be _some_ penalty. one that makes it sufficiently unattractive that people stop doing it so flagrantly.
Yeah. It was a god-spoof of the looney toons things where anvils would just randomly drop on people.
God's mighty anvil was great.
Something stupid/naughty would be going on "down on earth". Then you'd see a scene change to god walking around in a cloud, looking down, holding an anvil. he's squint downwards and line up the anvil just right, then drop it.
Then it would land on whoever was down on earth causing the issue of interest.
haha
:)
if you succeed in killing smss.exe, the machine goes away
similarly, if csrss.exe exits, smss.exe bluescreens the machine.
lsass is the local security agent subsystem server. (i always read this is "ls ass"
SMSS is the session management subsystem. it spawns Csrss.exe (Client Server Run Time SubSystem - the Win32 layer on top of NT)
If you have a suitably old smss.exe, it also spawns the OS/2 1.x layer or the POSIX layer. If you have Services for UNIX, there is a new posix.exe layer and psxrun.exe servers that you'll also see.
No you dont. Irix 6.3 was For O2's only. You're either running 5.x, 6.2, or 6.5 on your Indigo2. If its got the R8000 (power Indigo2) it might be running 6.01 or 6.1
Or, you have an O2, not an indigo2, in which case you're running 6.3 or 6.5
Well, i dont blame them. The 100mbit cards for the machines that didn't ship with 10/100 were expensive and SHITTY. For instance, for the indigo2, you could get an EISA phobos ethernet card which is basically a 3c597 with a different rom. This was spendy and you would never saturate the 100mbit card and it would generate a crapload of interrupts (slam cpu utilization).
Similar story for SGI Indy/Challenge S. The 100mbit workstation cards were 3rd party, spendy, and not so hot.
you could get SGI labeled 100mbit cards for the VME boxes (Challenge DM, for isntance) that worked pretty well.
well, noone will _make_ me apply that policy to my emails or images. Now i cant choose to do so if i want to. Tomorrow I may be able to. Palladium wouldn't restrict my ability to send email. It doesn't "restrict" anything - It lets ME restrict your ability to do things i dont want you doing with my stuff, which is _Exactly_ what I do want.
;) you're certainly free to release them in whatever unencumbered format you like. Palladium wont prevent that. It might even be beneficial to you.
The point is that the control of the content usage should be at the hands of its creators. The _only_ caveat is preserving fair use for copyrighted works. One advantage of big media companies being big media companies is if they categorically inhibit fair use by making bad DRM policies, they'll get spanked for it legally, eventually. The recent price fixing suit should be evidence enough that they're not the golden children of capitol hill.
I don't want you to have fair use of my emails, my photo album, or anything else thats mine. I want you to have exactly the usage that I say you have.
Maybe with palladium in place, the DMCA can be overturned ? Palladium doesn't attempt to restrict your ability to create software (DMCA does). It makes it technically hard (hopefully infeasable) to circumvent restrictions placed on digital information. Hopefully, legality will be removed from the picture altogether, which means dumb lawmakers wont be making dumb laws.
THat means that the dmca can go, and you'll be allowed to write your software dvd player. ANd if future dvds are released for palladium-only devices, too bad. Hollywood isn't obligated to release a DVD at all. When you're making better movies (which really wouldn't be much of a stretch, honestly
You're certainly free to develop a non-tcpa and non-palladium alternative for every widget in the stack, from cpu to web brwoser and everything inbetween. (unless fritz gets his way). you can't realistically get upset if nobody wants to use them besides you and a few slashdotters.
In a way, it's similar to the "this page doesn't work with opera on linux" argument. Sure, web designers probably _should_ test sites on every browser ever. But they dont. You have no right to tell them how to make their stupid site. You can just choose not to go there.
If you've tried visiting a website using Omniweb 2.x on NextStep, you know how useless much of the web is unless you're running a recent browser on a supported platform.
Ideological living is expensive and inconvenient. If palladium/tcpa have merits in the eyes of adopters(and it looks like thats a possibility), how can you plausibly be upset about its adoption ? Because it might inconvenience you ?
Please explain why you think that is the case.
you don't want everyone seeing your picture, encrypt it and send it to a friend.
If i encrypt a picture and send it to a friend, they simply unencrypt it and send it to some other "friend".
Without a comprehensive DRM scheme, there is no way to "give with restrictions". It's either give, or dont give.
Point taken about outputting the image to a DVI panel. You do know that there are provisions for anti-sniff DVI work, right ?
MS doesn't want to make an MS computer, because there's no money in hardware, and they know it. There is money in software. The xbox is a platform that puts MS in a good position to make lots of money on software - games.
I've had conversations with a few of the people actually working at MS on the drm/palladium stuff. It's funny that you're trying to tell me what MS's plans are in this regard.
Hrmm.
I for one am really looking forward to having a palladium infrastructure in place.
1) I wont buy shitty music regardless of how its encumbered, so that doesn't effect me.
2) DVDs are encumbered now, but i can do whatever i want to with them (watch them). There are black-market players that are available to me that let me skip commercials, undo region coding, etc, but as much as i dislike those attributes, i haven't voted with my dollars (yet)
3) If someone like microsoft (or apple, or sun, or any other technology company) outlines how DRM et al should work, that means record companies and legislators _ARENT_ doing it. When was the last time you ran across a good government standard ? When was the last time you saw a peice of software by a media company that was worth using ?
4) I really like the idea of being able to put a usage policy on content. For instance, if i scan some pictures of a vacation, and put them on a website, perhaps i want to apply a policy that says these can only be viewed by people that have access the picture from the URL. Then i can simply restrict access to the site via the normal means, and the pictures will not be redistributable once they've been downloaded by the viewers that I do allow.
you think this is stupid, and maybe it is. On the other hand, a picture taken of my wife and a friend of ours (totally clean, of course) from a party we went to ended up on a "hot teen of the day" site. Asking the site adminstrator to take the pciture down of course was a pain. And i was lucky that they complied.
If there had been some sort of DRM policy on JPEGs then i wouldn't have to be upset with the friend that posted the pictures on his site for not restricting access. I wouldn't have to flinch everytime i see a camera to think "where the fuck is this going to end up, maybe with a few photoshop edits"
The MS work behind TCPA/Palladium is not as orweillian as you think. It is not a framework designed to let hollywood hold all the strings (nor is it setup to let MS hold all the strings).
TCPA / Palladium do not add any restrictions. They allow new types of things to be done that cannot currently be done (sending an email that cannot be forwarded, distributing an image that cannot be edited, etc)
If media companies stop distributing media in clear-text, thats their choice. Vote with your dollars. If, despite rational arguments to the contrary, you cannot stand the idea of TCPA, you'll be able to vote with your dollars there, as well.
Bzzt.
The throttle in the E39 M5 and E46 M3 is electronic. They are drive-by-wire. There is no direct mechanical linkages.
Not that that matters. in my 88 BMW, i _do_ have a direct throttle cable (i've even replaced it), but i still don't need to retard timing to limit acceleration. The hard rev limiter is a fuel cut off. No timing modifications necessary. The momentary rev limit is a temporary fuel cut.
Naturally fuel cutoffs are probably too jerky for traction control applications, but they're definitely used for rev limiting. Most TC systems that need to actuate a slow-down condition use the brakes more than ECU work, however.
Actually i think the spark-retard thing has to do more with air temperature. If you read what Dinan has to say about their tuning on the E39 M5 package, they found that alot of the power loss comes from intake air temperature. The the ECU takes intake air temp as one of its inputs, and it knows that as the temperature of the air rises, it cant be as aggressive with its timing, so it slwos down. The biggest changes dinan made to the E39 M5 were on the intake, to keep the air as cool as possible. This lets the factory programming make the most of things while still staying inside the safety envelope.
Clearly running a car on the dyno thats _expecting_ 130mph of cool air coming at its radiator and intake is bad, and artificial air is a serious requirement.
I'm surprised that you say that the ECU retards the timing based on wheel speed. Many BMWs of course have traction control (and even the 80s bmws had 4 channel ABS) and clearly that cares about wheel speed differences. The TCU needs to talk to the ECU to tell it to cut throttle when wheel(s) spin, but it wouldn't do that with spark retard. Do you have any pointers to a discussion of the front-wheels-stationary problem ?