I am currently writing software for a client to replace his ancient Commodore PET system. It must be at least 20 years old.
The PET is still used nearly every day. It is set up to run a single program, which automatically calculates measurements for metal parts, glass, and screens for custom-built house windows, with an integrated "database" of window types. The software was custom-written. By today's standards, it is ugly, user-unfriendly, slow, and clumsy. But back then, it must have been a huge improvement over repetitive number crunching with calculators and pencils.
Any idea what a fully-functional Commodore PET system, with Commodore monitor and a HUGE external floppy drive cabinet (with two 5 1/4 inch drives) might be worth?
I have had significant problems with compatibility between NT and 95/98. These problems do exist.
I develop Win32 software professionally for a very small company using MS Dev Studio (C++, MFC, ActiveX controls). We aren't using anything like multimedia, DirectX, or any of the other newer API's.
A couple of examples:
1. Last year, I discovered that in the OnItemExpanding event for a MFC tree control, you can do a SortItems in Windows NT with no problem. But that causes a stack overflow on 95. (Worked around by using the OnItemExpanded event instead.)
2. I wrote an ActiveX control this year which draws into a bitmap and then bit-blits onto the screen. This dramatically reduces flicker. It works perfectly under NT, but under 95/98 the bitblt resulted in complete garbage on-screen. Never did figure that one out. (Worked around by detecting the OS at runtime and not using offscreen bitmaps unless it's NT. Sigh.)
Stuff like this is a nightmare to debug, and often only discovered near the end of the test-fix cycle.
Mere "proofs of existance" like that do exist for some problems. That is, there are proofs that polynomial time algorithms exist for some problems which do not actually show what the algorithm is.
Theoretical computer science (my former specialty) is a very strange field.
As far as I know, no commonly used encryption algorithms are based on NP complete problems.
However, if I remember correctly, it has been proved by Michael Fellows (a former professor of mine) that encryption algorithms can be created based on NP-complete problems, such that breaking the encryption can only be done in non-deterministic polynomial time (unless P == NP).
Sorry I don't have a reference here. If someone is really interested, I think I could find it. I don't remember if the encryption algorithm would be at all practical to actually use, however.
For those of us that have been following this for some time, the situation is pretty clear. I remember following the comp.org.eff newsgroup way back (1992?) when the Clipper chip concept was first introduced, and all these arguments were hashed out many times even then.
The US Government doesn't need to have breakable or escrowed encryption for investigating known criminals. In a criminal investigation, they can almost always get the information they need some other way - video or audio bugs, surveillance, TEMPEST, blackmail, force, etc.
But all those techniques are labor intensive. The whole point of escrowed or otherwise restricted cryptography is to reduce labor cost. That is the ONLY way to explain the US government's otherwise irrational encryption policy.
They want the ability to conduct easy, cheap, automated, wide spread electronic surveillance, Echelon-style, for all digital communication.
Once you understand this, all government encryption policy makes sense.
And of course, most guns in the US are never used against a human. So a person could, logically, be a "gun supporter" even if they would never condone using a gun against a human.
Another thing about Rambus memory: From what I have read, it has higher latencies than SDRAM. Apparently, this actually decreases performance in many situations.
No. A Celeron 300 is running a 66 Mhz bus with a 4.5 multiplier. These are commonly overclocked to 450 Mhz by increasing the bus speed. Intel has multiplier-locked all their recent chips, so increasing the bus speed is the only (easy?) way to overclock.
Interestingly, the Celeron 300A is more expensive locally than the 333 at the moment, because everyone knows about overclocking. (and most 333's won't run stably at 500).
Government regulation is not dumb and unneccessary.
It is easy to find out what kind of quality a software has by reading reviews of it. This is certainly true for games.
Two examples are "Sin" and "Unreal". Both are good games which were probably released a little too early. Their respective bugs and/or performance problems were extensively documented and discussed in on-line reviews within a week of their respective releases.
So what is the problem, other than dumb consumers who don't bother to find out what they are buying? That problem will never be fixed.
Those mission-critical ancient PC's had BETTER be replaced before the end of the year. They almost certainly have Y2K problems. This would probably have an impact on "data collection applications" or "process control".
ummm, actually those high-speed providers (at least where I live) won't give you a unique IP address. They use DHCP (?) at least partly because they don't want to undercut their expensive T1/T3 business-oriented services which DO give unique IP addresses.
What I am waiting to see, with some mixture of dread and fascination, is for the government to realize that some significant chunk of "economy" is taking place in these virtual worlds.
Then, of course, they will try to figure out how to tax it.
I predict that some national or state government will at least discuss taxing the current equivalents of "Ultima Online gold pieces" before 2004.
After all, if this stuff has real value, you could buy real things with it. Imagine if you could buy books from Amazon with your Ultima Online virtual gold.
The paper has dozens of grammatical errors. Cliches are everywhere. I didn't see any evidence of original thought.
Background research is nearly nonexistent. The author doesn't even refer to basic works in the field by Yourdon and others. In fact, only three actual books are mentioned in the bibliography, and that counts the New Hackers Dictionary! (Although "Microserfs" is quoted, it is not in the bibliography.)
That was truly "written as university research"? It would have barely rated a "B" as a Grade 7 report.
Slashdot Mirror
Wrong! Well, sort of.
Quake 1 and 2 did not support SMP.
Quake 3 does, on both Win NT and Linux.
Azog
You have a weird way of choosing what operating system to use.
Why don't you just use whatever works best for you, instead of using whatever OS has spokespersons you agree with?
And what about the "half-truths, omissions, and outright lies" told by the spokespersons for Windows, your current chosen OS?
How bizarre.
Azog
I am currently writing software for a client to replace his ancient Commodore PET system. It must be at least 20 years old.
The PET is still used nearly every day. It is set up to run a single program, which automatically calculates measurements for metal parts, glass, and screens for custom-built house windows, with an integrated "database" of window types. The software was custom-written. By today's standards, it is ugly, user-unfriendly, slow, and clumsy. But back then, it must have been a huge improvement over repetitive number crunching with calculators and pencils.
Any idea what a fully-functional Commodore PET system, with Commodore monitor and a HUGE external floppy drive cabinet (with two 5 1/4 inch drives) might be worth?
Should my client or I try to auction it off?
I have had significant problems with compatibility between NT and 95/98. These problems do exist.
I develop Win32 software professionally for a very small company using MS Dev Studio (C++, MFC, ActiveX controls). We aren't using anything like multimedia, DirectX, or any of the other newer API's.
A couple of examples:
1. Last year, I discovered that in the OnItemExpanding event for a MFC tree control, you can do a SortItems in Windows NT with no problem. But that causes a stack overflow on 95. (Worked around by using the OnItemExpanded event instead.)
2. I wrote an ActiveX control this year which draws into a bitmap and then bit-blits onto the screen. This dramatically reduces flicker. It works perfectly under NT, but under 95/98 the bitblt resulted in complete garbage on-screen. Never did figure that one out. (Worked around by detecting the OS at runtime and not using offscreen bitmaps unless it's NT. Sigh.)
Stuff like this is a nightmare to debug, and often only discovered near the end of the test-fix cycle.
Theoretical computer science (my former specialty) is a very strange field.
As far as I know, no commonly used encryption algorithms are based on NP complete problems.
However, if I remember correctly, it has been proved by Michael Fellows (a former professor of mine) that encryption algorithms can be created based on NP-complete problems, such that breaking the encryption can only be done in non-deterministic polynomial time (unless P == NP).
Sorry I don't have a reference here. If someone is really interested, I think I could find it. I don't remember if the encryption algorithm would be at all practical to actually use, however.
For those of us that have been following this for some time, the situation is pretty clear. I remember following the comp.org.eff newsgroup way back (1992?) when the Clipper chip concept was first introduced, and all these arguments were hashed out many times even then.
The US Government doesn't need to have breakable or escrowed encryption for investigating known criminals. In a criminal investigation, they can almost always get the information they need some other way - video or audio bugs, surveillance, TEMPEST, blackmail, force, etc.
But all those techniques are labor intensive. The whole point of escrowed or otherwise restricted cryptography is to reduce labor cost. That is the ONLY way to explain the US government's otherwise irrational encryption policy.
They want the ability to conduct easy, cheap, automated, wide spread electronic surveillance, Echelon-style, for all digital communication.
Once you understand this, all government encryption policy makes sense.
And of course, most guns in the US are never used against a human. So a person could, logically, be a "gun supporter" even if they would never condone using a gun against a human.
Yeah, all that stuff interferes with my FM radio reception too... you get the point.
Another thing about Rambus memory: From what I have read, it has higher latencies than SDRAM. Apparently, this actually decreases performance in many situations.
(Higher bandwith, but higher latency.)
Interestingly, the Celeron 300A is more expensive locally than the 333 at the moment, because everyone knows about overclocking. (and most 333's won't run stably at 500).
oops should have previewed.
Govt. regulation IS dumb and unneccessary.
Government regulation is not dumb and unneccessary.
It is easy to find out what kind of quality a software has by reading reviews of it. This is certainly true for games.
Two examples are "Sin" and "Unreal". Both are good games which were probably released a little too early. Their respective bugs and/or performance problems were extensively documented and discussed in on-line reviews within a week of their respective releases.
So what is the problem, other than dumb consumers who don't bother to find out what they are buying? That problem will never be fixed.
Chill out. Your hard drive and network card both have unique serial numbers that can be read by software.
The whole PSN thing is ridiculous, ignore it.
Those mission-critical ancient PC's had BETTER be replaced before the end of the year. They almost certainly have Y2K problems. This would probably have an impact on "data collection applications" or "process control".
ummm, actually those high-speed providers (at least where I live) won't give you a unique IP address. They use DHCP (?) at least partly because they don't want to undercut their expensive T1/T3 business-oriented services which DO give unique IP addresses.
What I am waiting to see, with some mixture of dread and fascination, is for the government to realize that some significant chunk of "economy" is taking place in these virtual worlds.
Then, of course, they will try to figure out how to tax it.
I predict that some national or state government will at least discuss taxing the current equivalents of "Ultima Online gold pieces" before 2004.
After all, if this stuff has real value, you could buy real things with it. Imagine if you could buy books from Amazon with your Ultima Online virtual gold.
No, it's not "just service", because it can be traded amongst users. That is the point that you seem to have missed.
Most "services" like cable TV, etc. are not possible to exchange between users.
The paper has dozens of grammatical errors. Cliches are everywhere. I didn't see any evidence of original thought.
Background research is nearly nonexistent. The author doesn't even refer to basic works in the field by Yourdon and others. In fact, only three actual books are mentioned in the bibliography, and that counts the New Hackers Dictionary! (Although "Microserfs" is quoted, it is not in the bibliography.)
That was truly "written as university research"?
It would have barely rated a "B" as a Grade 7 report.
Faugh.