Matrox & Creative Labs = a couple HALF-ASS driver packages!

[merchant2112]

hey i vote for deleteing this one.

didn't he just violate a few roc's?

[PaulS]

Yeah right, it's all Microsoft's fault. Do the self appointed hardware apologists actually have any facts to back up their assertion that W2k is buggy?

So how much slack should we cut the vendors? 6 months you reckon? Did it ever occur to you that if people didn't expect their drivers to work properly for the first 6 months then el slacko hardware vendors wouldn't start working on them until a month before they were supposed to be working? I mean, w2k has only been in development for a couple of years right...

Paul

[Himself]

LOL.

[spoogenet]

Gotta love the "buggy software" arguments.

A) (not to sound like a MSFT employee) The 63k bugs is "unconfirmed" by the company officially. Sure that means a lot, I just wnated to get it out in the open so everybody can start ranting about that.

Now on to the more important things:

1) Nothing is 100% perfect. I'm sure your nice 0 bug tolerance in hardware is great, but have you ever thought of reading Intel's list of errata for processors? If you run an Intel processor, then you certainly don't adhere to your 0 bug policy.

2) Software is extremely complex and also RELIES ON HARDWARE! Yes, that's right boys and girls. What sometimes appears to be a software glitch is actually a hardware bug. Since you can't necessarily get a new piece of hardware to replace the one you've got, then software patches are issued to fix (if possible) or work around the problem.

Since the software is so dependent upon the hardware then it is very easy to mistake a hardware bug for a software one. Everybody is quick to blame MSFT (not to support them) when something goes wrong without considering other possibilities. OS programmers have quite a task at hand to make an attempt to deal with all the problems.

Since we have now considered the possibility of >0 bugs in hardware, let's look more into the matter. As an OS programmer, one must try to make the software to work with as much hardware as possible. This often causes problems such as needing a certain BIOS version or higher (read W2K's Knowledge Base about Adaptec SCSI cards) and such. Sounds to me like Adaptec doesn't make 0 bug hardware afterall. That aside, go shopping one day around the net and look at all the different pieces of hardware, their revisions, manufacturers, etc. I guarantee you the list will be very long. Now go back and write an OS that is compatable with ALL of these. That's right, if you want 100%, 0 bugs, then it must work with every single one of these.

That's quite a task to perform or to ask of somebody. 0 bugs in software is near impossible with something as grand as an OS.

Try to keep things in perspective and you may realize that what you want is rather difficult to get. If you want to pay $5000 for the consumer version of Windows, then you may get almost bug free software, but I'm sure it'll have at least 1 bug in it.

So why don't we all get off the "bash MSFT" wagon and maybe try to be practical about it. NO OS is bug free, and probably no hardware is bug free either.

And remember, boys and girls, it's not a bug, it's a feature!! )

[PaulS]

If there is no chip design software for NT, it would be purely a marketing decision. The only exception may be applications that require more than 8 processors (prior to Windows 2000 NT didn't scale that well). There is nothing about NT that makes unsuitable for engineering applications.

I'm not trying to let anybody off the hook for buggy software, but Windows 2000 is probably the most stable piece of software (for it's size) ever written.

Hell, I wish Matrox could write drivers as stable as their hardware. Maybe they should sack their programmers and give the work to their chip designers.

Paul

[Wombat]

PaulS: We do put entire OS's on chips, they're called embedded systems. If they're not right the first time, there's usually not an upgrade. If there is, it ain't easy. And don't comment on how much code is in "silicon" if you haven't seen the code.

Thunderchez: What did the Intel guys expect you to use for software? We use Mentor Graphics here, and it's the biggest reason we have our Sun boxes.

[HollyBerri]

Software is extremely complex and also RELIES ON HARDWARE! Yes, that's right boys and girls. What sometimes appears to be a software glitch is actually a hardware bug.

And sometimes what appears to be a video glitch is a memory glitch, a soundcard glitch, or a mobo glitch. We've all seen it.

People seem to forget 2 things:[list=1][*] Everything in a computer system is interdependent, if only to the extent that any given piece of software relies on all the others to do as they've said, release memory as promised, etc. Naturally, most everything relies on everything else much more than that.[*] Everything that happens is eventually displayed through the video subsystem, even if the process doesn't involve the video subsystem... so any errors will involve the video in some way, whether or not there's anything actually wrong with the video card or drivers.[/list=a]

So games may run slow or not at all because the sound card did not initialize... but what you will see is a black screen (because sound initializes first--thinking of DOS games-- and then video... and there was no message, or it went by too fast. So you think there's something wrong with your video card, but there isn't...

But, seriously, you guys-- isn't the problem just what has been said all along... that with so much variety in HW configs (and consequently, in drivers installed), not to mention in software configs (tray apps, disabled Windows components, old 16-bit apps that you just can't give up, config utilities and more) that it is really impossible to eliminate all conflicts (which seems to be what most "bugs" boil down to in the end) from the piece of SW you (as a driver programmer, or a game programmer, or a HW utility programmer) have written, or must write?

I kinda think we're lucky to have everything work so well so much of the time... but then I'm amazed that there are so (relatively) few auto accidents, that the traffic laws actually stem the tide of anarchy....

--------------------
Holly

[PaulS]

Wombat, wrong conversation. I was not talking about silicon as another storage medium. I have worked on embedded systems, you design compile and test the code in an emulator and then burn it to some kind of static memory. Big deal, embedded systems can only be embedded because they are tiny. How many embedded systems do you know of the exceed 20 million lines of code and take several hundred meg to store?

Paul

[SCompRacer]

Get a dictionary and learn some new words to express anger or frustration without profanity.

[Thundrchez]

OK, this is digressing a bit.

Intel went through a big kick a few years ago to "design Intel on Intel". It is just mildly ironic that if you get a job at Intel as a chip designer that you sat in front of a Sun workstation.

Intel then pushed Cadence to have a winnt port of their tools. I wonder if they are actually using them.

There is nothing about NT that makes unsuitable for engineering applications.

There are plenty of reasons not to use winnt as a ic design platform. The sysadmin burden is about 50X as much as we have for our Suns, with the right automated setup. Try sending a batch of 30 circuit simulations to run on 30 workstations in winnt. Oh, that's right, you can't remotely log into a NT box. Perhaps you need your process to run on the big server, but how do you get it to display back to where you sit? Oh, that's right, no remote displays either. Windows (any flavor) is fundamentally setup for only one person to sit in front of the box and be the only person logged into it.

Yes, hardware can have bugs in it. We should set the goal to be 0 bugs. Hardware manufacturers go through very stringent design and testing to make it that way. The standard should be no different from software. We should not let ourselves be brainwashed into saying that bugs are acceptable in software. If you need a refresher course, please install win3.0 or win95a on a system. Microsoft may be getting better, but we fundamentally cannot accept bugs in software as normal. Software bugs should be found and fixed.

[spoogenet]

I don't think the issue is all of us being "brainwashed" into accepting bugs. My main point is that we can't all just blame the software for every problems that arises. Also we can't realistically expect every piece of software to be compatable with every other piece of software and at the same time be compatable with all the possible hardware configurations.

Ever think of the actual number of hardware configurations that could exist? Now make your software 100% compatable with all of it. Along the lines of what I have already said and what HollyBerri was saying, computers are complex webs of interdependency. For it to be 100% bug free, 0 bugs, EVERYTHING has to work perfectly.

A lot of problems that MSFT faces in its programming is trying to make it work when sloppy programmers come along and mess up other applications. If problems occur, MSFT is blamed rather than the sloppy programmer who actually caused the problem. Kinda like blaming the car for the wreck, rather than the person driving it.

I don't think any of us really likes bugs in software, but I can certainly understand how and why they exist. Now if EVERYONE would just do their best to make 0 bug things, then OS programmers would have a much easier job.

And it's also nice that everybody is really quick to point out when the system fails, but not that many people actually pay any attention to when it works. Maybe we should give people more credit when it works and they'd be more inclined to make it work.

'nuf said.

[ahartman]

Is part of the problem, maybe, that software makers are catering to too large of an audience?

Is there really any reason that Q3 needs to be written so that it can run on a Pentium processor? How playable is Q3 on a P1 anyway?

Why does Windows (95, 98, Me) still have DOS? Because people still run DOS programs.

It wouldn't be the most popular thing, but if you remove support for older technologies, you have less code and therefore (I believe anyway) less bugs not to mention more time to improve the newer technologies...

Would anyone do that? No way. Cuts out too much of a potential purchaser base.

[Gurm]

Jammrock:

I'm a bit offended that you think of me as the designated defender of Microsoft. I simply have arrived at the conclusion that there is no LOGICAL reason to be upset with Microsoft TECHNOLOGICALLY. Their business model sucks wind. Hardcore. But their technology has always been very well-intentioned, if nothing else.

Others:

I think the issue we're overlooking here is that the current driver situation is EVERYONE'S fault.

1. Matrox is to blame for not devoting enough time to their drivers... EVER. Sometimes I honestly think that there's one guy in a room somewhere being whipped by a stern taskmaster who shouts "code the driver! code the driver!"... and that's the whole development team. Other times I'm less cynical. But Matrox is SLOOOOOOOW to develop drivers, as we all know.

2. Microsoft is to blame for their mixed-goal approach to Win2k. If it's a business OS, then go with that. If not, then fess up and support games in word as well as in deed. Cut it out with this "oh, it's for business, so it's ok that no games run, but we've included DirectX7, and here are some game bug fixes!" nonsense.

3. Game developers are to blame for cutting corners. The same issues that make each new video card prone to game glitches make Win2k prone to game glitches. If the coders were to use standard Win32 and DirectX code to begin with, they wouldn't have this problem. Look at well-written games... Quake3 needed no patches to run under Win2k. It has since been updated, but not for that reason.

4. Thunderchez is to blame for thinking that writing software is ANYTHING like making silicon. SHAME ON YOU. Go write an OS (I have, have you?) and then come back and tell us how easy it is to kill bugs. Ok? For that matter, all of you need to go write a video driver (once again, I have done this and doubt many or any of you have). Then come back and complain about Matrox's drivers.

- Gurm


------------------
Listen up, you primitive screwheads! See this? This is my BOOMSTICK! Etc. etc.

[Thundrchez]

Gurm, I'm the one who said that writing software is the same as designing silicon? Perhaps you missed this from one of my posts regarding just that...

that is an apples to oranges comparison. In crude theory you might could argue such a comparison, but it does not go far.

People who write software do not test their code anywhere near as much as people who design silicon test their design. It is very obvious, IMHO. And because software is not tested anywhere near as much as it could or should be, we get buggy software. If you want to have low standards for software, enjoy the bugs. But it does not have to be that way.

[SCompRacer]

In this increasingly complicated, high tech we-try-to-satisfy everyone society, is it really possible for companies to suspend shipment of any product until it is "bug free?"

This really is a point of view issue that can be based on one's level of expertise. An end user can say "this does not work well, their programming is horrible." A programmer can say it with conviction.

In these times, it seems perfection in anything has been replaced by cash flow to sustain business or maintain market saturation.

--------------------------------------------
"When you are waist deep in hungry alligators, it is hard to remember your primary objective was to drain the swamp."

[This message has been edited by SCompRacer (edited 24 March 2000).]