What other hardware/Technology is on the horizon?

[DemoCoder]

Ah, good old slashdot, where Microsoft bashing is not just in vogue, but virtually required in every single damned thread, even if the news story is about lamb in Australia. Where people routinely claim that the average Microsoft NT installation crashes multiple times per day.

Funny how anyone who actually works in a corporate environment (running servers or help desk) hasn't had this experience. Linux zealots actually think that when the X-Server or buggy Linux clone apps (KDE office, Gnome, etc) core dump, the fact that they don't take down the whole OS and spill you back to the command line where you have lost all of your work is a huge consolation to former Windows users. Where not having to reboot the kernel, but reboot every other GUI process in the system is a significant advancement in the history of computer science. I'm sure my moment knows how to run GDB on the core and strip out some of her lost documents.

You know, Microsoft is not stopping you from writing, let's say, a killer email application or browser. They really aren't. Opera software, for example, wrote a really nice browser and is doing well in the mobile/PDA market. For PC's they tried to charge money or make you watch a permanent AD banner in the browser. Most people don't want to deal with this if a non-spyware-enabled free browser exists. The fact that consumers won't buy a web browser is not really Microsoft's fault. I would never by a web browser, even if Microsoft didn't exist. Before Netscape, I was using Mosaic and Viola already. And I expect plenty of "free" browsers would have came along and comoditized the market already.

You're just going to have to get used to the fact that you can't write one piece of software and just keep selling the same thing over and over again. If you want your business to survive, you've got to keep adding to it, giving people a reason to upgrade or to purchase new versions of it.


Once a piece of software becomes very widely deployed (let's say GIF decompression algorithm), people expect it to be there by default, like the air your breathe, and you can no longer make a viable business say, selling GIF viewers. Very quickly, if there is any money in GIF viewers, the market will be flooded by them, the price will be reduced to near ZERO quickly (because the marginal cost of copying information is next to zero) by all the competition, most of the companies selling GIF viewers will go out of business, and at that point in time, GIF decompression has been rendered a commodity.


I'm sorry if you think Microsoft is evil for bundling utilities and applications into the OS distribution which are not longer differentiable products, but it is good for the vast majority of consumers. And Microsoft software doesn't suck anymore than buggy crap that you download RPMs for Linux.

It's rose colored glasses than sensitizes people to any mention of a Windows bug, by blinds them to any Unix bugs. Just sit in bugtraq or security-focus mailing lists. Daily there is some new security exploit for Unices, but it's not major news. The moment someone finds even the most trivial Windows exploit, it's all over Slashdot as "proof" how bad Microsoft programmers are. If you get a Kernel panic on Linux, someone will just tell you that you have a bad kernel module or distribution, but if it happens on Windows? Oh boy, more proof that Windows NT/XP is inherently unreliable written by suck-ass programmers.

 

[Kristof]

(1)Is there any compatiable problem when some campany use the techology like tiler?
(2) Can we make the Z-compression ratio up to 1: 10 instead now 1:4 and result in huge performance increase?

Not entirely sure if I understand the questions.

I suppose (1) is about compatibility of tile based rendering, there should be no compatibility problems. If your worried about KYRO not running all games anymore then its most likely due to hardcoded need for TnL or other features like Cube Mapping.

Not sure what (2) is about, the thing to remember is that if compression has to be lossless you have no control over how much you can compress. If you allow lossless there is a huge risk for artefacts which these days in not acceptable. And IMHO scenes get more complex (more objects, smaller triangles, complexer objects, noise in the VS, ...) hence Z buffer contents gets more complex and hence compression efficiency will get worse over time.

 

[Fuz]

Actually AFAIR there has been public comments from IMGTEC themselves that Series 5 will be released next year one way or another (as in if they find a partner or not).

Thanks, I would love to see a release sooner rather than later.

I am very interested in whether the power VR will be marketed from now on as a contender for the performance crown, or as strictly a value proposition for the low end of the 3D OEM market.

Same here, I would prefer if they go after the high end.

But in the end we'll just have to prove it

See, maybe I wasn't dreaming!

 

[Ailuros]

They have a hellavu lot to proove if you consider the gap between Series3 and Series5

 

[JF_Aidan_Pryde]

Democoder: Australian lamb rocks.

 

[MfA]

Windows (except NT/2K) is inherently less secure because as a rule you run everything with administrator privileges. Maybe they will sidestep that issue in the future completely, with security being such a focus point, by using something a little more finegrained than ACLs. Frankly I hope not though, I would find it amusing to see macro virus's wreak havoc on a palladium "secured" system.

 

[madshi]

The pascal language isn't the best for an experienced programmer though IMO.

Hey Humus, I don't agree at all. Look at all the buffer overrun problems in Windows/IE. IMO they are home made problems because of using C(++). The Delphi compiler can check your code much better. Still you can produce buffer overruns, but it's much more difficult than in C(++). I'm doing low level OS system hacking with Delphi (e.g. see http://help.madshi.net/madCodeHook.htm ), and it's absolutely fun doing that with object pascal. I see no reason why pascal shouldn't be just perfect for both beginners and expert programmers.

P.S: Slightly off topic, sorry...

 

[PC-Engine]

They have a hellavu lot to proove if you consider the gap between Series3 and Series5

Well they had already started development on Series 4 a while ago, but then when STM halted development of Series 4 some of that design was move over to the MBX series. Series 5 I think will probably benefit from the development of Series 4 so it shouldn't involve a lot of catch up work.

 

[Tahir2]

Shocking but we use Win98 Network Edition (eh? thats what i thought).

Its not entirely stable but then again its not always crashing, about the same as NT from my user experience.

Oh and its fast and zippy even on modest hardware... still weird using win98 for 400 odd networked PC's

noticed this:
Democoder

Many of the same people who whine amount Microsoft including a DVD Player in XP (puts WinDVD/PowerDVD out of business) didn't say a word when Apple included a DVD player built into the operating system.

What DVD player?
You need a DVD pack and that needs to be purchased with WinXP at least and Windows MediaPlayer. MS even does recommend other DVD playing software inc, PowerDVD when you click on the appropriate prompts the first time you put in a DVD movie.

demalion:
But that's just nitpicking.. I will say if MS products were not QUALITY or fitted the needs of the public then they wouldnt sell.... alternative or no alternative.

Regarding PowerVR Series 5.. me can't wait

 

[KimB]

demalion:
But that's just nitpicking.. I will say if MS products were not QUALITY or fitted the needs of the public then they wouldnt sell.... alternative or no alternative.

I disagree. With such a huge installed-base, MS products really can be made pretty darned crappy and still do very well. I think most people end up purchasing MS products just because it's easy to do, not necessarily because they're better.

One quick example is MS Frontpage. Put simply, Frontpage can really produce some incredibly buggy webpages (often that will work with IE, not Netscape). I loaded a Frontpage-created webpage into another web page designer, HotMetal, and it immediately detected and fixed the bugs.

From what I can see, MS's products tend to be easy to use, easy to learn, but quite buggy, and usually not as fast as other products once you get to know how to learn them.

 

[Humus]

The pascal language isn't the best for an experienced programmer though IMO.

Hey Humus, I don't agree at all. Look at all the buffer overrun problems in Windows/IE. IMO they are home made problems because of using C(++). The Delphi compiler can check your code much better. Still you can produce buffer overruns, but it's much more difficult than in C(++). I'm doing low level OS system hacking with Delphi (e.g. see http://help.madshi.net/madCodeHook.htm ), and it's absolutely fun doing that with object pascal. I see no reason why pascal shouldn't be just perfect for both beginners and expert programmers.

P.S: Slightly off topic, sorry...

An experienced programmer shouldn't buffer overrun. The unsafe function calls are wellknown and should be avoided when security is critical.
What I meant though is that the language itself can be quite cumbersome and you get quite verbose code even for simple things.

 

[DemoCoder]

Both DreamWeaver and Mozilla can also produce buggy web pages, so what's your point? Producing HTML from a WYSIWYG Editor that works on multiple browsers and multiple screen resolutions with non-buggy layout is a very hard problem.

Most human beings when hand editing HTML will produce buggy HTML unless they agressively test against all browsers and use helper tools. One of the reasons why XHTML had to be introduced was to put some rigor back into the syntax, since most browsers were so forgiving (unbalanced quotes, spurious tags, and overlapping tags aren't errors and most of the time the browser will fill in the missing details)

Again, we have selective evidence. HotMetal probably has its own set of bugs. It isn't exactly the #1 editor, is it?


Mfa,
Inherent less secure than Unix which has no fine grained security in the kernel at all?

Not everything runs as administrator, only most of the system services (inetd, network file system, security services, etc) which is the same situation as Unix. Sure, NT has more stuff running as admin compared to Unix, then again, Unix doesn't have fine grained security and has had far more exploits than NT against it.

For example, on Unix, the only fine grained security constraint you have is filesystem permissions. There is no way to assert that a user cannot open a socket, for example. Either you are superuser and can call any Unix function (e.g. open ports < 1024), or not. This forces lots of wrapper-launcher programs which start as root, acquire resources, then fork a child process that is the real meat of the program. However, about 90% of the people who write such wrappers make mistakes and end up opening up a race condition that allows an exploit.

A large number of security exploits would have been prevented if you can simply assert that a process has no ability to open new sockets (for remote shells), fork new processes, etc -- a more sandbox oriented model. Currently, the only way to sandbox a process is to CHROOT() it, but that only sandboxes the filesystem, it won't prevent you from using the machine as a DDOS system.

The only reason Unix seems to be inherently more secure is that #1 it is alot older #2 it was running on the inherent as attackable servers far longer than NT #3 because of widespread exploits against it over the past 10 years, code has been systematically hacked and audited to be more secure.

But even after all these years since the Morris Worm took down tens of thousands of Unix servers, there are still buffer overflow exploits in today's unices. Even DNS BIND was found to have another exploit in the last month!

There is nothing, architecturally better about the Unix kernel security wise tho. It's all because of the massive patching and fixing of all the scripts and daemons written over the past ten years. For all the work, it's still a house of cards IMHO. One bug, one lazy programmer, and your Unix server is wide open again.

 

[MfA]

As I said, except NT/2K ... Ill admit it was a strange way of putting it, but then most people do not run NT/2K and when there are rants about windows's (lack of) security it is mostly aimed at those people running 95/98/ME/XP (and implicitly those people who run everything as root under nix .

Marco

PS. buffer overflows are so lame, I think bound overruns in general could have been made impossible at low cost a long time ago through some simple changes to computer architecture and languages ... I blame the present situation on Real Programmers and C.

 

[madshi]

An experienced programmer shouldn't buffer overrun.

Heh, so that means that all the security holes in Windows were produced by non experienced programmers? So then Microsoft must have a whole bunch of them!

What I meant though is that the language itself can be quite cumbersome and you get quite verbose code even for simple things.

Some examples, please?

Well, I don't want this to get a C vs. pascal holy war. Probably it's a matter of taste. I prefer clear and readable source code over a cryptical sequence of characters (what C looks to me, sometimes). For sure both languages have their strong and weak points. I fail to see, though, what this has to do with which language fits better for experienced programmers.

 

[Althornin]

As I said, except NT/2K ... Ill admit it was a strange way of putting it, but then most people do not run NT/2K and when there are rants about windows's (lack of) security it is mostly aimed at those people running 95/98/ME/XP (and implicitly those people who run everything as root under nix .

I fail to see how you lump windows XP in with win9x, etc?
Everything you run in XP does not run as administrator.

 

[KimB]

Again, we have selective evidence. HotMetal probably has its own set of bugs. It isn't exactly the #1 editor, is it?

Its primary problem is functionality. It has essentially no support for scripting. As you were saying, many webpages "fill in" the missing details. This is one of the reasons I was very impressed with HotMetal. It did a very good job at checking the syntax. This doesn't seem like a challenging thing in the least to do, WYSYWIG or not. After all, the problems you were describing are simply related to browsers being entirely too forgiving. Just make the editor have rigorous tests, and you'll be okay.

In other words, I can understand scripting/Java problems in webpages across different browsers. Basic functionality is another thing, and there shouldn't be an editor in existence that produces errors with basic functionality. But, apparently, there is. As a side note, I never liked Netscape's editor (Never tried Mozilla's).

The way I see it, there are humungous problems with most software development today in checking the basics. I think it has to do with a lack of planning software development, and a lack of checking to ensure the basics in the software (ex. that each function does what it is supposed to, that each resource is used at it was designed to be used, etc.).

Granted, I will agree that Microsoft isn't the only offender in bad coding (otherwise there'd be no need for patches for nearly every piece of software in existence!). I'm mostly upset with them for their business practices (primarily forcing OEM's to sign contracts that force them to ship Windows with every PC sold, and not allowing the uninstallation of certain Windows features).

 

[MfA]

In XP the standard user account has most of the same priviliges.

 

[KimB]

I'd also like to just go ahead and add in my little two cents about Linux vs. Windows.

Currently, the biggest problem with Windows is with regards to interdependencies. It really seems like every piece of Microsoft software depends on other pieces. This can make it very challenging to uninstall specific pieces of software. There are also silly little interdependencies. As an example, installing IE5 actually changed some API things, such as how CD autorun works. This is just a silly correlation to me, and one thing among many that makes me upset with Microsoft.

As for Linux (I've never used Unix, so I really don't know...), there are a number of things that I really like, but also a number of drawbacks. The drawbacks with Linux pretty much all lead back to the Linux user needing to know a heck of a lot more about how the system works in order to get Linux to work. One of the primary things this stems from is the open-source nature of Linux, as well as the vast variety of distributions available. This has created two primary problems: a lack of a unified driver interface (which often makes vendor-specific drivers not very easy to install), and the inability to run many programs without recompiling from the source code on the target machine.

The lack of a unified driver interface can certainly make driver debugging in Linux much more challenging. The frequent need to ship source code with software (esp. drivers, but also a fair amount of higher-level software) certainly requires more work for the user, and may have also turned off many would-be Linux developers from the platform.

So, the way I see it, I like Linux, to a point. It really needs some work, though. Hopefully United Linux will help to resolve many of these issues, but I don't think that they can all be resolved without making it at least somewhat proprietary (ala OSX). Personally, I don't really like the Macintosh UI, but it really could be, in theory, pretty much everything I want Linux to be.

 

[Basic]

Not that I have that strong of an opinion here, but I think this link fits in rather well in the discussion:

http://www.despair.com/limitations.html

Or maybe not, but at least there's a lot of funny stuff there.

 

[DemoCoder]

The interdependencies exist in Windows because Microsoft is way more aggressive about using objects and reusing code. Before widespread usage of shared libaries on Unix, typically the runtime was completely statically linked for each app. Even as shared libraries started to creep in, they mostly remained core libraries.

However, if you've downloaded the latest version of Linux, from say, Mandrake or Redhat, and try to install an upgraded version of a library or app, you'll get dozens of comments from RPM telling you to upgrade all the other dependencies. So Linux is moving in the direction of Windows.

As they rely more and more on reusable system level objects, and apps export their objects for use by other apps, you will see more interdependencies. KDE and KDE Office are great examples of this on Unix.