|
Once support for a hardware peripheral is created, it should exist forever for all future versions of that Operating System.
Designing changes into Operating Systems changes (versions, upgrades, revisions, call-it-what-you-like) that are not compatible with existing driver methodology is the problem.
I would expect support for peripherals to be different switching to a new O.S. - from Linux to Windows, Windows to Mac, Mac to OS/2, Windows to OS/400, Linux to Unix, for examples. But Windows x to Windows y, or Linux a to Linux b - no hardware peripherals should become outdated.
The fact that they do, or that they require significant work from somebody (hardware vendor), just indicates a poor design of the software driver support in the first place. Why is it tolerated? Don't be so darn short-sighted or lazy in the driver integration design, I say.
None-the-less, 10 years is obviously not long enough.
Gary
|
|
|
|
|
ghle wrote: The fact that they do, or that they require significant work from somebody (hardware vendor), just indicates a poor design of the software driver support in the first place. Why is it tolerated? Don't be so darn short-sighted or lazy in the driver integration design, I say.
Why do you think mainframes cost millions? Just as one example, the IBM channel technology allows magnetic drums and 4-foot-wide 2-megabyte Winchester drives to run beside state-of-the-art disk arrays. On the other hand, why do you think PC operating systems are so cheap?? You get what you pay for, and no one who lives on PC operating systems wants to pay for the kind of reliability and backward compatibility that the IBM channel architecture provides.
No, I don't work for IBM, and I don't program for mainframes 
|
|
|
|
|
cpkilekofp wrote: Why do you think mainframes cost millions? ...you get what you pay for,
I think mainframes cost millions because they get what you pay for. Supply and demand.
If I got what I pay for, Gates and his buddies would not be worth gadzillions, and the world (you and me) would not waste untold gadzillions explaining, fixing, supporting, working around Windows bugs! Just how many billion $$$ do you think MS takes in for, say, Windows XP? Less than IBM for their OS?
My point is that once a driver is written for an OS, it should be no work to use that SAME driver in a newer version of the OS. No rewrite necessary - minimal changes at best.
cpkilekofp wrote: no one who lives on PC operating systems wants to pay for the kind of reliability and backward compatibility
I would think my hospital would want this kind of reliability. I sure want them to have it! I think I want my government to have it, too.
I pay peanuts for that new Interstate bridge, but I sure don't expect it to fall down and collapse within a year. [edit] I also expect it to support any car I'll ever drive, including a 20-year old junker. [/edit]
Maybe if PC programmers learned to think, design and program for mainframes first we wouldn't have (pick any one) security issues, driver issues, OS crash issues, UI issues, speed issues, configuration issues, usability issues, and legacy issues. Rather than after-thoughts, things would be built right in the first place. That would be cool for all of us. 
Gary
modified on Monday, March 17, 2008 11:02 AM
|
|
|
|
|
ghle wrote: I think mainframes cost millions because they get what you pay for. Supply and demand.
Spot on. Most people just don't get this simple concept.
ghle wrote: Just how many billion $$$ do you think MS takes in for, say, Windows XP?
It's amazing how many people think MS doesn't overcharge when they are like the 2nd richest company in the world. Wal-mart only beats them out in sheer volume - not markup.
I always find it humorous that Apple - with their pricey machines - still manage to sell their OS for a third of the price and their office suite for a lot less and still make do.
|
|
|
|
|
ghle wrote: Maybe if PC programmers learned to think, design and program for mainframes first we wouldn't have (pick any one) security issues, driver issues, OS crash issues, UI issues, speed issues, configuration issues, usability issues, and legacy issues. Rather than after-thoughts, things would be built right in the first place. That would be cool for all of us.
*nods* I've heard that from mainframe programmers. I've asked more than one of them about their delivery schedules. After I stopped laughing, I explained to them that the product cycle of many PC components (hardware and software) didn't last that long. They humphed and harrumphed and said it SHOULD last that long. I gently told them that any PC shop that TOOK that long risked losing their market share, and thus their ability to pay for that sort of product cycle.
Your PC is NOT a mainframe. It's a much less reliable, much less expensive gadget the original design of which was crabbed together for a market which IBM thought would never exceed 250,000 products in the field. Still, it delivers functionality and ease of use that, for the tasks for which it was designed, no mainframe can match.
I could go on in this vein for quite some time...but then, I've actually done the analytical work to understand the differences, why they existed, and what should be done about them. But then, this is the Lounge, where working through the issue takes a distant second place to having a good time So taking you to the woodshed for your incorrect, insulting, and downright silly generalization would be too much work 
|
|
|
|
|
cpkilekofp wrote:
*nods* I've heard that from mainframe programmers.
I've never coded for a mainframe, however, I learned, like you, the whys and wherefores of doing such.
Your criteria seems to be this perversion that says because PCs are gadgets, they shouldn't be designed to operate reliably. The "code it fast and therefore less reliable" mentality - along with the myth that no one can create bug-free software - perpetuates this mess we have. Ouch.
I'd go to the woodshed with you to have a good time, but evidently like your analytical software decisions, that "would be too much work". 
I contend that more thought process up front - in the O.S. - could alleviate a lot of this problem of outdated drivers. You can't likely convince me otherwise. We know that Microsoft, for one, makes design decisions for the benefit of hardware suppliers. Whether the driver design is deficient by design or by laziness will have to wait for disclosure at some later time. 
Gary
|
|
|
|
|
You misunderstand, and I mean completely. You also have given no evidence of any understanding of complexity theory and its application to large scale software development.
ghle wrote: Your criteria seems to be this perversion that says because PCs are gadgets, they shouldn't be designed to operate reliably. The "code it fast and therefore less reliable" mentality - along with the myth that no one can create bug-free software - perpetuates this mess we have. Ouch.
You've obviously never heard of the famous "300". That would be the number of bugs left in the IBM 360 operating system after several service releases, say around 1970. They started with 1,000 plus bugs, got it down to 300...and after that, every fix they made spawned a new bug or set of bugs. It was TOO COMPLEX to functionally isolate every potential failure case and fix it without creating a new failure case.
ghle wrote: contend that more thought process up front - in the O.S. - could alleviate a lot of this problem of outdated drivers. You can't likely convince me otherwise. We know that Microsoft, for one, makes design decisions for the benefit of hardware suppliers. Whether the driver design is deficient by design or by laziness will have to wait for disclosure at some later time.
You mean like actually using the Mach kernel paradigm, as a for-instance? Get real. They're not in the market to make the best operating system, they're in the market to make MONEY. Lots of it. So is virtually every software and hardware entrepeneur out there. MS knew when they were developing OS/2 that Mach provided a better model, the entire INDUSTRY knew it, but they didn't see that it would pay them to go that route (despite their talk at the time of leveraging the Mach paradigm in their future systems).
ghle wrote: I'd go to the woodshed with you to have a good time, but evidently like your analytical software decisions, that "would be too much work".
Now you're just being nasty as well as naive. Programmers don't make those decisions. Business people do. Bill Gates didn't build DOS from the ground up as a PC OS; he bought an OS, hired a programmer to modify it enough so he could sell it, then he sold it. If it would have been cheaper and quicker to write one from scratch, he'd have done it. If his company had been convinced no one would buy Vista unless it was fully backward compatible, they'd have built it that way.
Reliability is expensive. Backward compatibility is expensive. I believe (despite your snobby, sophomoric, and ultimately ignorant comment) that both are worth the price if you value your customers' good will. I've put lots of effort to back that opinion when it could have (and, occasionally, did) cost me dearly. It's always been worth it to here the guy who got promoted instead of me tell me "you were right, we should have done it that way, can you do it that way now?"
However, I was pointing out why MS didn't do it, not telling you why I wouldn't have done it. For some more insight into why MS didn't do it, check out Joel on Software's latest blog entry. A genius of IT such as yourself should be able to find it easily.
|
|
|
|
|
cpkilekofp wrote: You misunderstand,
What I do understand is that forcing hardware manufacturers into constantly redeveloping software drivers to interface to a new version of an operating system is one indication of a bad design of an operating system. Are people rich because of that mentality? Sure, but that doesn't make it right.
Pick up "Dreaming in Code" by Scott Rosenberg. I found this outtake appropriate to the topic.
"... it is enough to dream of adding another new layer to today's stack of turtles. [Read Stephen Hawking if you don't understand the turtle reference.] That's how the field has always advanced in the past, they point out, so why not keep going? But there are other critics of today's software universe whose diagnosis is more sweeping. In their view, the problem is the stack itself. The brittleness comes from our dependence on a pile of unreliable concepts. The entire field of programming took one or more wrong turns in the past. Now, they say, it's time to start over."
I say, do it right in the first place, then we wouldn't have this mess. Since we have the mess, fix it. Your reference to the browser standards problem highlights the chaos that is created when it's not done properly. (No ethics discussion on whether MS put specific quirks into their code so web sites would break if IE was not used. Just ask et al Novell for their opinion.)
Now what path to take? Upset the apple-cart because MS-specific applications will now fail, or put up with more crap and even extend it's negative impact? Maybe reverse the table on MS - if the site doesn't work, let MS pay to fix it. It's like Y2K all over again, isn't it. But the reasons are different.
Continue along the path of ship-it-quick and make some bucks, and support it over and over and over again while profits decline - I care not. I just want my 2-month old camera to work with a newly released OS. I want my 3 year old printer, 5 year old scanner, and 8 year old hard drive to work, also. Profits for Windows, lost profits for every one else, but that's decision that business people make, I guess.
"You makes your choices and you pays the price." - George Kostell
Gary
|
|
|
|
|
cpkilekofp wrote: Your PC is NOT a mainframe. It's a much less reliable, much less expensive gadget
Misguided assumption, I think.
"...superior price/performance; greater reliability, availability and serviceability..." and "Linux 2.6 kernel offers an array of enterprise and performance features"
http://www-03.ibm.com/systems/linux/power/marenostrum/hpc/index.html[^]
Gary
|
|
|
|
|
ghle wrote: Designing changes into Operating Systems changes (versions, upgrades, revisions, call-it-what-you-like) that are not compatible with existing driver methodology is the problem.
I'll have to disagree with ya on that one. Take Microsoft's Vista (as this survey was obviously inspired by Vista and its driver problems...)
Vista has a re-engineered model for almost anything. Drivers, GUI, TCP/IP, Audio, anything. I do not know how much has changed, but MS is clearly saying that they've re-implemented all these. What this means in practice, is that they've removed (or, at least tried to remove) the problems that plagued the old driver model, etc. Vista can really be thought of as a "new OS". So much has changed since previous versions of Windows, that it is simple not worth keeping the old code and compatibility hacks around in the system.
What this means for device manufacturers, is that they have to adapt their hardware for the new model (or, rather adapt the drivers). Granted, its more work, and results in only the newer hardware being given Vista support, but that's, really, stupidity on the manufacturers' end.
ghle wrote: But Windows x to Windows y, or Linux a to Linux b - no hardware peripherals should become outdated.
They don't become "outdated". They still do their jobs fine, its just that the manufacturers all of a sudden decide to ride the "oooh, all the more reason to push new pricey stuff out" wave when a new version of an OS comes out. In reality, there are absolutely NO problems writing drivers for every single piece of hardware that, say, WinXP supports in order to make them compatible with Vista. Seriously. The diver to the hardware is sorta like the GUI to an application - it communicates information to the user, and gathers input for him. Only the "user" in this case is the OS.
The *nix world is even worse - manufacturers either don't know _how_ to write drivers for those systems, or they simple chose to ignore the much smaller market.
|
|
|
|
|
Anton Afanasyev wrote: Vista can really be thought of as a "new OS".
I agree somewhat, but I think 95/98 to NT/2000 was a bigger delta for a "new OS".
With the 95->NT, or XP->Vista conversion, the opportunity to start anew and correct the driver DLL hell were laid at their feet, to correct mistakes originally made. So far, it's still broke.
Anton Afanasyev wrote: They don't become "outdated".
I disagree here. If the peripherals don't work anymore, then they are outdated for that O.S. Sure, the hardware still works if you have the old OS running somewhere, but for the purposes of the discussion, the peripherals are now worthless - they don't work. They need updated. Now, if the manufacturer creates a driver for the new O.S. the updated driver will un-outdate the peripheral.
Anton Afanasyev wrote: In reality, there are absolutely NO problems writing drivers for every single piece of hardware that ...WinXP supports in order to make them compatible with Vista.
Well, this is the gist of my problem. Every manufacturer should not have to write a driver for a new version of an OS. The new OS comes out, the old driver should work fine. Maybe not with any whiz-bang capabilities the new OS might offer, but they should work. Think of the cancel button on the GUI. I can run my old app in Vista, but it has the fixed-size X button unless I rewrite the app because there was no way to size it differently from the others. (Yeah, I know there's a "way", but that's not the point here.)
If MS came out and said, "we're sorry, but Vista requires a redesign of drivers, but you'll never have to do it again" that would be a much better story to tell than "we're sorry, but Vista requires a rewrite of the drivers just like last time and next time."
But, that's just my opinion.
Gary
|
|
|
|
|
ghle wrote: Designing changes into Operating Systems changes (versions, upgrades, revisions, call-it-what-you-like) that are not compatible with existing driver methodology is the problem.
But hardware manufacturers are also to blame here. There was an old driver model, and many drivers didn't follow the rules, but somehow just worked. A new driver model comes, with backward compatibility for good old drivers, and those bad old drivers cease to work. Whose fault is it?
MS goes to great lengths to support old software in newer versions of Windows. Maybe that is why Windows is so bloated; it has just too many compatibility hacks. Have you ever read Raymond Chen's blog[^]? He sometimes explains compatibility hacks, and the great length MS goes to ensure an old program that doesn't follow rules still works.
|
|
|
|
|
|
Its time for codeproject community to be the third party 
|
|
|
|
|
Given the amount of suckage in the usual HP software (think supermassive black hole at the center of our galaxy), why would you want it anyway?
Software Zen: delete this;
|
|
|
|
|
Gary Wheeler wrote: Given the amount of suckage in the usual HP software (think supermassive black hole at the center of our galaxy), why would you want it anyway?
So we could add enough stuff to eventually fill up the black hole?
Gary
|
|
|
|
|
I'm seriously disillusioned with pissed at HP at the moment.
I've owned several HP printers, all of which have been fantastic from a hardware perspective. They're reliable, last practically forever, and aren't too painful on consumables. Their driver software has always been crap, but the Windows builtin support worked well.
That was true until my most recent purchase. I bought an HP PhotoSmart CS5240. It's a scanner/printer combination.
The software was the usual HP nightmare. At the end of the installation, it had two HP Photosmart printers in the my device list. The first one was disabled and offline. The second was enabled, but only for photo tray printing. After an hour of cussing, multiple reboots, and strong words with the Device Manager, I got that mess straightened out. I could finally print.
Good Lord, what a piece of noisy junk! This thing sounds like it's a Borg with epilepsy. Printing a page requires 30 seconds of clanking before it even gets paper into the printer. Print quality is OK, but not awe-inspiring. The thing sucks ink like an alcoholic. My daughter's printed a little over a dozen 4"x6" photographs, and the color cartridge is empty. I figure the thing costs around $1.50 per photograph, which is ridiculous.
At the moment I don't think I will ever buy another HP product. I'm stuck with this one, since I can't get a refund based solely on "poor performance".
Software Zen: delete this;
|
|
|
|
|
You would get along very well with Rajesh. Maybe you can form a club.
Cheers,
Vikram.
Zeppelin's law: In any Soapbox discussion involving Stan Shannon, the probability of the term "leftist" or "Marxist" appearing approaches 1 monotonically.
Harris' addendum: I think you meant "monotonously".
Martin's second addendum: Jeffersonian... I think that should at least get a mention.
|
|
|
|
|
If they drop driver support, they should at least make the driver source code available so we can develop drivers on our own for new OS's.
As far as support, if the device is still in production (or still in the retail channel) when the "new" version of Windows comes out, they should have drivers for it.
Lastly, avoid buying the multi-function printers. You're always going to be faced with driver problems regarding one or more functions of the printer.
"Why don't you tie a kerosene-soaked rag around your ankles so the ants won't climb up and eat your candy ass..." - Dale Earnhardt, 1997
-----
"...the staggering layers of obscenity in your statement make it a work of art on so many levels." - Jason Jystad, 10/26/2001
|
|
|
|
|
Until the time that they clearly mention when selling the product. However no one bothers identifying it in advanced. What prevents any company to say that I'll support you for 5 years, for example?
// "In the end it's a little boy expressing himself." Yanni
while (I_am_alive)
{
cout<<"I love to do more than just programming.";
}
|
|
|
|
|
Hamed has hit the nail on the head. Make a commitment and stick to it. That's what my company does. We invested a huge amount of effort in updating drivers for Vista for products we don't even sell any more. With a clear commitment, customers can decide for themselves.
And this business of asking a bunch of enterprise programmers to make a marketing decision for peripherals manufacturers? Downright silly. Of course they're all going to say "a damned long time" because anybody who hasn't written a mass-market driver for a non-trivial device thinks it's easy.
|
|
|
|
|
Sure it's not easy and I don't defend the thesis of supporting it forever but at least 2 generations!
We see hardware coming out not compatible with Vista, only XP and this is not acceptable but if the hardware was bought to Windows ME  I don't demand it to be supported on Vista.
Another problem I know it happens is that companies don't want the drivers to last too long because they want to sell the hardware. Not supporting old hardware on new OS's is the easiest way to "force" a customer to upgrade its piece.
|
|
|
|
|
You mean my HP550C might not work under Visa?
But it ran under 3.1, 95, 98SE, and XP Pro, and is only 15 years old!
Shame on you HP!
Bob Emmett
|
|
|
|
|
Bob Emmett wrote: You mean my HP550C might not work under Visa?
Not one of those H-1B Visas, anyhow. http://en.wikipedia.org/wiki/H-1B_visa[^]
Seriously, I did expect a 2-month old HP digital camera to work with the new HP Vista computer. After tech instructions to reload the camera OS, new (free under warrantee) USB cables, reload the OS, untold hours on phone with tech support, they finally owned up that the camera, even in USB external drive mode, was not compatible with Vista. That's inexcusable.
Gary
|
|
|
|
|
Maybe so, but it's an entirely different issue.
|
|
|
|
|
General 
News 
Suggestion 
Question 
Bug 
Answer 
Joke 
Praise 
Rant 
Admin 
Submit an article or tip
