The time stamp counter is not used for keeping time, dude, only the number of clock cycles since bootup.
http://www.cs.wm.edu/~kearns/001lab.d/rdtsc.html
Secondly, your premise that changing the software compiler language will save the world is a load of foo. CPU's running all code in kernel space would bring us back to the dark ages in terms of protection against hacks and security of data to name a few.
I suppose you think that Java will save the day? Little do you know how much java depents on native code execution (JNI). Parts of java are written in C, for instance. Just look up java security flaws in 2003 on google for an overview.
And yes, in memory mapped IO, the MMU controls access to the bus. When a call to an illegal memory location is detected (for instance, by a malfunctioning driver) the MMU will trigger an exception on the processor. Since most devices use a mixture of MMIO and interrupt driven drivers this is easily handled.
Removing the mmu and bounds checking every memory access would cause your shiney new opteron to run at much reduced speeds. To try this out, run a few complicated Java Swing applications.
Listen, I'm not disagreeing that "We need to change software." What we need despirately is better testing tools, languages that scrutinize inputs from foreign sources such as user, network, and disk. Compilers that really analyze not just optimize.
I'm open to new ideas, but I'm tired of cookie cutter development tools. It's good to see things as a child in this respect, to look at what should "just happen"
Thanks for your insight, and don't give up. It's time to revolutionize software. The advent of HTML brought millions of people into the computing world, and it's ripte for a fresh change.
heh.. reading time generally means going out on the bus (The south bridge in PCI architecture) and hitting a timer chip. Because this means a lot of latch/unlatch on the bus, plus cache coherency there's going to be a hit. Try reading random locations from memory.. say on your AGP card, and you'll find the same issues.
Not a function of your C compiler, try again.
As for a safe language, that is a nice utopian ideal, but thank goodness that MS didn't write windows in it. Can you imagine all the Virus and trojan horses? Memory management is a must. What about the case where a removable device failes to work because of a bad contact? With an MMU this is fairly trivial to kill the whole process and cleanup the bus. With a safe language, there would be an enormous amount of exceptions which would have to individually handle many different cases.
However, I will completely agree with your last point.. software can be a lot more powerful, usable, and safer than it is. Whoever invents it is going to be a hero!
I know what you're saying, but they're not telling him to switch to Mac because it's a better operating system, they're telling him to switch because these applications are only available on Mac.
I'm writing to share a tragic little story.
My Dad has a PC that my sister and I used to use for our DVD's. One night, I was writing a DVD on it, when all of a sudden it went berserk, the screen started flashing, and the whole DVD just disappeared. All of it. And it was a good DVD! I had to cram and rewrite it really quickly. Needless to say, my rushed DVD wasn't nearly as good, and I blame that PC for the grade I got.
I'm happy to report that my sister and I now share an Apple PowerBook. It's a lot nicer to work on than my dad's PC was, it hasn't let me down once, and my grades have all been really good.
IANAL, however the GPL depends on the copyright law, and copyright owners certainly have the right to revoke licenses when the license is broken.
Distributing software which you have no license too is definetely going to get you nailed. In this case, SCO will have to prove that they still acknoweledge the GPL license, or they will get nailed big time in copyright court. Since they have written Congress that the GPL is invalid I don't see how they can get out of this.
I call BS!! I seriously DOUBT they sold 100,000 CPU's last year.. prove it.
"For the calendar year 2002, they sold around 3,700 processors - compared to over 479,000 total 64-bit RISC processors. That's a market share of only.007 or.7%,"
"In the first quarter of 2003, sales volume slipped 31% to 1,963 units resulting in only $63 million of revenue. A quarter-to-quarter sale decline of this magnitude for a product that is 'gaining momentum' (according to HP/Intel) is not a positive sign."
I can't find any more real info after that.. just Gartner and others "predicting"
We're not talking cells here, or random DNA, choices, privacy, or science. Save that for the lab. Let's also forget about religeon here.
We're talking about humanity. We're talking about the greatest walking, thinking, talking creation known! Life is mysterious and undefinable because it is personal. Why do you love? Why do you care about your Mom, Sister or Dad? What makes superstars so attractive? Why is someone cute? Because they have a value or property that is rare, special, and intimate. This value is what makes it sacred over an animal, plant, or piece of land. An ordered person knows the value of person over things.
It is sacred because it is one's greatest gift. If someone was to take your life it would be the ultimate theft of your potential deeds, happiness, and companionship. Inside we know this, everyone has this innate sense. Show a child the act of murder and you WILL see a reaction to death that's fearful.
What is a race that kills its own child? The greatest potential a human achieves? It's cannibalism of the body. Barbarism of the value of a man or woman. If life is cheap, then you and I are worthless, expendable and ultimately simply workers to feed the rich, unsatiable, and flesh hungry.
How is C or Perl any different in functionality from Java in terms of freedom? C and Perl still require fluff to initialize and "do stuff."
In terms of complexity, I don't think the answer lies in the language. Look at the System 370 OS and tell me that's not complicated. Most of it is in assembler and machine language. Yet IBM wrote and supported the system for numerous generations of hardware and OS design. They have CPU partitioning, transparent multiple IO busses. They even ported Java, Perl, and C to it! Complexity is expensive to maintain in the brainpower department though.
Freedom isn't the operator here.. it's money. Reducing complexity means that cheaper and less educated labor can work in smaller chunks of logic which have strictly defined interfaces. Inferior solutions can be replaced when required and are easier to drop in.
That's not to say that Java is for dummies, as many well educated smart people are it's biggest fans! But business has driven java into the mainstream. It did the same thing for C. Otherwise we'd all be using C++ now, right?
Back to Java though, if you look at the features provided, they are all aimed at reducing error prone code through mitigating stupid errors.
Examples include strict compilation rules, strict naming conventions, security errors are removed by fencing the programmer away from pointers. The list goes on to include sockets, threads, memory management, etc.
This combines to form an interesting cost analysis that finally results in one cost that scares businesses more than anything!
RISK!
With a good C or Perl solution, the risk is higher that a programmer will make a mistake that can't be fixed with more hardware, bandwidth, or programmers. With Java, mistakes can more often be glazed over with these things. This reduces the risk and makes business happy.
To think that programmers are driving any of this other than in the academic sense is comic in a sense. Getting a room full of programmers to standardize on anything is an excercise in futility, even in the Java community.;) When Bob the CIO comes in and mandates Java - the discussion is over.
I often thought of working collab.net when it first appeared... the problem was the amount of buerocracy involved in the system, the amount of work put up by the developer which at any time could fail, and the whole process starts over.. Here's my thoughts:
1. More along the lines of slashdot would work better, with moderators scoring out the good and the bad.
2. Then a pot of sponsers can be setup once the issue closes.
3. After sponsership closes, then a functioal specification is written by potential developers.
4. Once a spec is accepted by a quorum of sponsers, then the developer goes and implements.
5. If he fails then others are allows to pickup where he left off, or the pot can be shelved.
The buerocracy of collab.net put me off the first time I read it... this is open source, and should be more relaxed, IMHO.
People want to have a hand in creating something great.. they just need an opportunity!
Troll, I feed you!!...But installing mozilla, for multiple users on windows? Find the win32 intaller, click on it.
Go to mozilla.org, get install RPM, save to file. Find the Home icon on your desktop, go to the mozilla installer RPM and click on it. For all users, install as root (Sound familiar?) BTW, these instruction refer to the gnome desktop.. I don't use KDE all that much.
Try to find where redhat installed mozilla, since a simple find feature it appearently something linux doesn't need.
The find utility is located on the red hat menu, 4th item from the bottom on the main menu, properly named "Search for files..." If you absolutely must know all the files mozilla installs, go to command line and type rpm -ql "app name" (app names is the first part of the file name, so mozilla would do the trick in your case)
If you want to uninstall mozilla (which does leave all your bookmarks in place) just do RedHat->System Settings->Add/Remove applications. Click on "Graphical Internet" applications, and unselect mozilla. Uninstalled. Your plugins will probably not work - as they will be targeted at your old install anyway (conflicts can occur on any platform)
To install an app RPM, just locate the RPM, go into Nautilus file manager and double click on it.
When you encounter a plug-in requirement on a page, mozilla will happilly install it using the handy dandy auto-installer as long as the page properly references the installer (Most do these days)
My uptime is in the hundred day range since I last rebooted redhat. I'd check the system logs (Redhat->System Tools->System Logs) for possible errors you might be encountering. More than likely it is a hardware problem.
Don't go blaming your unfamiliarity as system design faults.. doing low level stuff on windows requires the same type of tweaking (registery editing, dll registry, ini file editing, etc...)
Actually, IIRC, you can modify your own DVD player to skip commercials.. you just can't tell anyone how you did it.:P
Pan
Re:...and the .NET Framework is language-neutral
on
Does C# Measure Up?
·
· Score: 1
Obviously, you've never been through a merger.. or a series of mergers.
Personally, I prefer to keep the platforms rolling as they are than try to port all code back to the company standard. Things like messageing and database unification can often solve the worst problems.
Well, I've been working on this exact problem for a long time now, as have numerous others. I am a (currently lurking) member of Xfree86 (back in the dark days when you had to sign non-disclosures). I've met and talked with a few of the core people even.
Here are my suggestions for XFree86: 1. Simplify the server.. let me guess that 90% of the code is redundent, out of date, etc. Really, a nice re-organization of the codebase would make it a lot more coherent. The framebuffer rewrite got me excited, but lets keep going. A basic tree might look like this:/include/server/lib/server/communication/server/protocol/server/modules/server/windowing/server/rendering/DRI/server/rendering/Mesa/server/rendering/XRender/server/font/server/drivers/xxxx/server/plugins//client/lib/client/blah blah
You get the idea.
2. Get rid of font servers. Seriously, integrate font management into X. I mean adding and removing fonts from the server at the user level too.
3. Replace the base rendering model with XRender (or allow a mixture). It's time some of the extenstions moved into the core server (Shape anyone??)
4. Let the server cache graphics list. This will help abstract gtk and qt toolkits from the rendering. That way, a server can be loaded with a description of a button, and take care of the drawing and refresh of that button. I'm _not_ talking NeWS here.. I'm talking "what graphic primitives redraw this component." These lists could be shared between KDE and GNOME. You could create them in SVG and they could be translated to X primitives by the toolkits. Then, toolkits only need to manage a single SVG file. Wanna new look for your desktop, just drop in a new SVG.
5. Modularize the core. Ouch, that will hurt.. but sometimes people want to use X just for a device setup and a framebuffer. (Think embedded). Re-architect around the idea that X is an orchestrator of devices, inputs, and graphics primitives. That was the original spirit of X, and should carry on.
6. With all that in mind, kill imake. Seriously, who uses imake besides X? Bueler? Bueler?
7. Clean up Xlib. Merge the other libs into the library. We have smart linkers these days, ya know. Since we killed imake, we can use configure or something along those lines to fix this.
8. Document it all. Document how a window is created, and what parts there are all the way down to the rectangle lists. How this list is translated into graphics onto a screen. XAA is fairly well documented. XVideo is a bit rough. XRender has somewhat real documentation, and you can read the thoughts of the designers on the public lists.
Even Plato acknowledged a warrior class! It's simply not possible for a Utopia of this nature here. Whatever your moral arguments for or against Iraq, it is here and now. You want soldiers on your side in this situation.
Yes, there are programs that do this. One interesting thing is fixed floating point. It's basically where you say one integer register can hold enough significant digits to emulate a floating point in limited situations (games, some rendering types, etc)
The big advantage is that integer operations tend to get ganged through the ALU at a high rate (and on some cpu's simultaniously.. i.e. alpha).
This could also be useful for brute-force cracking on RSA, etc..
Anyhow.. YMMV, limited liability, and always seek legel advice before investing in risky ventures.
Direct3d's API works across all video platforms so long as the drivers are compliant. No extensions, so you write your code and it works on all cards. Nonsupported features are emulated in software.
Bit of a herring there.. OpenGL was the first HW assisted 3D graphics API to actually support software fallbacks. Any "compliant driver" can be designed to do the same, whether in OpenGL or D3D.
Direct3d is meant for what we have today instead of being the same set of commands which have been extended over and over again.
This is only through the wonders of COM.. Which also means a lot of your CPU time is spent doing COM translations. I would argue though that having a C++ or some object oriented API for OpenGL would be wonderous.
"Comparisons of languages, libraries, and programming styles are notoriously tricky, so please do not draw sweeping conclusions from these simple tests."...
Hrm.. I've built a system that does this using 4x redundancy, though for security products not trading.;-)
We were able to scale 128 million transactions per hour on 16 PIII-500 boxes. The key to our transactions were shallow FIFO queues. Before transactions were executed, the queue list would require certain "sub transactions" to have been performed on 4 separate, though "unreliable" systems. Only 2 transactions were required for completion, and at that stage we executed the transaction and responded. Failed or timed-out transactions would be de-queued and responded to.
The key to speed was shallow (one level deep) message queues, and only guaranteeing 50% of the sub-transactions. Since each component was redundant, once 2 were completed we had full verification of the security operation. When the transaction commenced, we marked the transaction ID as "in use" and all other sub-transactions were tossed.
Another interesting tidbit on our system was the up-front design requirement for multiple data centers. So all the "sub-transactions" were actually performed on systems in CA, TX, IL, and VA. That's why they're "unreliable" (and latency-prone as well).
Also, store and forward transactions were distributed across the 4 locations, with the requirement that transactions could not be lost. Since our data was stored securely, we couldn't just simply keep a copy in a file or table, so we stored the Transaction ID's into a primary transaction engine (system which controlled top level interface to outside systems, somewhat like a transaction switch). When system load fell below certain percentages, the transaction engine would execute recommit commands to the data storage switches and build new keys for the data based on timestamps, and then recommit them back to all the data storage systems.
Our biggest performance hurdle was the fact that security algorithms are very slow to work with. The best way to cope with 2-3 second transactions was through massive parallelism of the transactions, so that even though all transactions were 2 or 3 seconds, we could perform a thousand per second per location.
Slashdot Mirror
The time stamp counter is not used for keeping time, dude, only the number of clock cycles since bootup.
l
http://www.cs.wm.edu/~kearns/001lab.d/rdtsc.htm
Secondly, your premise that changing the software compiler language will save the world is a load of foo. CPU's running all code in kernel space would bring us back to the dark ages in terms of protection against hacks and security of data to name a few.
I suppose you think that Java will save the day? Little do you know how much java depents on native code execution (JNI). Parts of java are written in C, for instance. Just look up java security flaws in 2003 on google for an overview.
And yes, in memory mapped IO, the MMU controls access to the bus. When a call to an illegal memory location is detected (for instance, by a malfunctioning driver) the MMU will trigger an exception on the processor. Since most devices use a mixture of MMIO and interrupt driven drivers this is easily handled.
Removing the mmu and bounds checking every memory access would cause your shiney new opteron to run at much reduced speeds. To try this out, run a few complicated Java Swing applications.
Listen, I'm not disagreeing that "We need to change software." What we need despirately is better testing tools, languages that scrutinize inputs from foreign sources such as user, network, and disk. Compilers that really analyze not just optimize.
I'm open to new ideas, but I'm tired of cookie cutter development tools. It's good to see things as a child in this respect, to look at what should "just happen"
Thanks for your insight, and don't give up. It's time to revolutionize software. The advent of HTML brought millions of people into the computing world, and it's ripte for a fresh change.
Just sit on the ad page, and after 2 or three seconds it takes you to the static content ad. But you have to have javascript turned on too.
Pan
heh.. reading time generally means going out on the bus (The south bridge in PCI architecture) and hitting a timer chip. Because this means a lot of latch/unlatch on the bus, plus cache coherency there's going to be a hit. Try reading random locations from memory.. say on your AGP card, and you'll find the same issues.
Not a function of your C compiler, try again.
As for a safe language, that is a nice utopian ideal, but thank goodness that MS didn't write windows in it. Can you imagine all the Virus and trojan horses? Memory management is a must. What about the case where a removable device failes to work because of a bad contact? With an MMU this is fairly trivial to kill the whole process and cleanup the bus. With a safe language, there would be an enormous amount of exceptions which would have to individually handle many different cases.
However, I will completely agree with your last point.. software can be a lot more powerful, usable, and safer than it is. Whoever invents it is going to be a hero!
Pan
I know what you're saying, but they're not telling him to switch to Mac because it's a better operating system, they're telling him to switch because these applications are only available on Mac.
I'm writing to share a tragic little story.
My Dad has a PC that my sister and I used to use for our DVD's. One night, I was writing a DVD on it, when all of a sudden it went berserk, the screen started flashing, and the whole DVD just disappeared. All of it. And it was a good DVD! I had to cram and rewrite it really quickly. Needless to say, my rushed DVD wasn't nearly as good, and I blame that PC for the grade I got.
I'm happy to report that my sister and I now share an Apple PowerBook. It's a lot nicer to work on than my dad's PC was, it hasn't let me down once, and my grades have all been really good.
Just switch!
IANAL, however the GPL depends on the copyright law, and copyright owners certainly have the right to revoke licenses when the license is broken.
Distributing software which you have no license too is definetely going to get you nailed. In this case, SCO will have to prove that they still acknoweledge the GPL license, or they will get nailed big time in copyright court. Since they have written Congress that the GPL is invalid I don't see how they can get out of this.
See the fun about too happen?
Pan
Don't forget that IBM supplied the machines and punch cards to process Nazi prisoners. Truly technology at it's worst.
Pan
I call BS!! I seriously DOUBT they sold 100,000 CPU's last year.. prove it.
.007 or .7%,"
"For the calendar year 2002, they sold around 3,700 processors - compared to over 479,000 total 64-bit RISC processors. That's a market share of only
"In the first quarter of 2003, sales volume slipped 31% to 1,963 units resulting in only $63 million of revenue. A quarter-to-quarter sale decline of this magnitude for a product that is 'gaining momentum' (according to HP/Intel) is not a positive sign."
I can't find any more real info after that.. just Gartner and others "predicting"
Pan
Absolutely true.
We're not talking cells here, or random DNA, choices, privacy, or science. Save that for the lab. Let's also forget about religeon here.
We're talking about humanity. We're talking about the greatest walking, thinking, talking creation known! Life is mysterious and undefinable because it is personal. Why do you love? Why do you care about your Mom, Sister or Dad? What makes superstars so attractive? Why is someone cute? Because they have a value or property that is rare, special, and intimate. This value is what makes it sacred over an animal, plant, or piece of land. An ordered person knows the value of person over things.
It is sacred because it is one's greatest gift. If someone was to take your life it would be the ultimate theft of your potential deeds, happiness, and companionship. Inside we know this, everyone has this innate sense. Show a child the act of murder and you WILL see a reaction to death that's fearful.
What is a race that kills its own child? The greatest potential a human achieves? It's cannibalism of the body. Barbarism of the value of a man or woman. If life is cheap, then you and I are worthless, expendable and ultimately simply workers to feed the rich, unsatiable, and flesh hungry.
Pan
That's truly an interesting thought to me..
;) When Bob the CIO comes in and mandates Java - the discussion is over.
How is C or Perl any different in functionality from Java in terms of freedom? C and Perl still require fluff to initialize and "do stuff."
In terms of complexity, I don't think the answer lies in the language. Look at the System 370 OS and tell me that's not complicated. Most of it is in assembler and machine language. Yet IBM wrote and supported the system for numerous generations of hardware and OS design. They have CPU partitioning, transparent multiple IO busses. They even ported Java, Perl, and C to it! Complexity is expensive to maintain in the brainpower department though.
Freedom isn't the operator here.. it's money. Reducing complexity means that cheaper and less educated labor can work in smaller chunks of logic which have strictly defined interfaces. Inferior solutions can be replaced when required and are easier to drop in.
That's not to say that Java is for dummies, as many well educated smart people are it's biggest fans! But business has driven java into the mainstream. It did the same thing for C. Otherwise we'd all be using C++ now, right?
Back to Java though, if you look at the features provided, they are all aimed at reducing error prone code through mitigating stupid errors.
Examples include strict compilation rules, strict naming conventions, security errors are removed by fencing the programmer away from pointers. The list goes on to include sockets, threads, memory management, etc.
This combines to form an interesting cost analysis that finally results in one cost that scares businesses more than anything!
RISK!
With a good C or Perl solution, the risk is higher that a programmer will make a mistake that can't be fixed with more hardware, bandwidth, or programmers. With Java, mistakes can more often be glazed over with these things. This reduces the risk and makes business happy.
To think that programmers are driving any of this other than in the academic sense is comic in a sense. Getting a room full of programmers to standardize on anything is an excercise in futility, even in the Java community.
Pan
Well, those rainbow and ncipher cards are great general purpose crypto cards but they don't help much when doing number sieving etc..
pan
I often thought of working collab.net when it first appeared... the problem was the amount of buerocracy involved in the system, the amount of work put up by the developer which at any time could fail, and the whole process starts over.. Here's my thoughts:
1. More along the lines of slashdot would work better, with moderators scoring out the good and the bad.
2. Then a pot of sponsers can be setup once the issue closes.
3. After sponsership closes, then a functioal specification is written by potential developers.
4. Once a spec is accepted by a quorum of sponsers, then the developer goes and implements.
5. If he fails then others are allows to pickup where he left off, or the pot can be shelved.
The buerocracy of collab.net put me off the first time I read it... this is open source, and should be more relaxed, IMHO.
People want to have a hand in creating something great.. they just need an opportunity!
Just a few thoughts..
Pan
Troll, I feed you!! ...But installing mozilla, for multiple users on windows? Find the win32 intaller, click on it.
Go to mozilla.org, get install RPM, save to file. Find the Home icon on your desktop, go to the mozilla installer RPM and click on it. For all users, install as root (Sound familiar?) BTW, these instruction refer to the gnome desktop.. I don't use KDE all that much.
Try to find where redhat installed mozilla, since a simple find feature it appearently something linux doesn't need.
The find utility is located on the red hat menu, 4th item from the bottom on the main menu, properly named "Search for files..." If you absolutely must know all the files mozilla installs, go to command line and type rpm -ql "app name" (app names is the first part of the file name, so mozilla would do the trick in your case)
If you want to uninstall mozilla (which does leave all your bookmarks in place) just do RedHat->System Settings->Add/Remove applications. Click on "Graphical Internet" applications, and unselect mozilla. Uninstalled. Your plugins will probably not work - as they will be targeted at your old install anyway (conflicts can occur on any platform)
To install an app RPM, just locate the RPM, go into Nautilus file manager and double click on it.
When you encounter a plug-in requirement on a page, mozilla will happilly install it using the handy dandy auto-installer as long as the page properly references the installer (Most do these days)
My uptime is in the hundred day range since I last rebooted redhat. I'd check the system logs (Redhat->System Tools->System Logs) for possible errors you might be encountering. More than likely it is a hardware problem.
Don't go blaming your unfamiliarity as system design faults.. doing low level stuff on windows requires the same type of tweaking (registery editing, dll registry, ini file editing, etc...)
Pan
Actually, IIRC, you can modify your own DVD player to skip commercials.. you just can't tell anyone how you did it. :P
Pan
Obviously, you've never been through a merger.. or a series of mergers.
Personally, I prefer to keep the platforms rolling as they are than try to port all code back to the company standard. Things like messageing and database unification can often solve the worst problems.
Pan
Some hacker took over your box and made someone with a phat line mad.
Pan
I remember this a bit as well.. A Siemens chip?? Anyone have more information?
Pan
Well, I've been working on this exact problem for a long time now, as have numerous others. I am a (currently lurking) member of Xfree86 (back in the dark days when you had to sign non-disclosures). I've met and talked with a few of the core people even.
/include /server/lib /server/communication /server/protocol /server/modules /server/windowing /server/rendering/DRI /server/rendering/Mesa /server/rendering/XRender /server/font /server/drivers/xxxx /server/plugins/ /client/lib /client/blah blah
Here are my suggestions for XFree86:
1. Simplify the server.. let me guess that 90% of the code is redundent, out of date, etc. Really, a nice re-organization of the codebase would make it a lot more coherent. The framebuffer rewrite got me excited, but lets keep going. A basic tree might look like this:
You get the idea.
2. Get rid of font servers. Seriously, integrate font management into X. I mean adding and removing fonts from the server at the user level too.
3. Replace the base rendering model with XRender (or allow a mixture). It's time some of the extenstions moved into the core server (Shape anyone??)
4. Let the server cache graphics list. This will help abstract gtk and qt toolkits from the rendering. That way, a server can be loaded with a description of a button, and take care of the drawing and refresh of that button. I'm _not_ talking NeWS here.. I'm talking "what graphic primitives redraw this component." These lists could be shared between KDE and GNOME. You could create them in SVG and they could be translated to X primitives by the toolkits. Then, toolkits only need to manage a single SVG file. Wanna new look for your desktop, just drop in a new SVG.
5. Modularize the core. Ouch, that will hurt.. but sometimes people want to use X just for a device setup and a framebuffer. (Think embedded). Re-architect around the idea that X is an orchestrator of devices, inputs, and graphics primitives. That was the original spirit of X, and should carry on.
6. With all that in mind, kill imake. Seriously, who uses imake besides X? Bueler? Bueler?
7. Clean up Xlib. Merge the other libs into the library. We have smart linkers these days, ya know. Since we killed imake, we can use configure or something along those lines to fix this.
8. Document it all. Document how a window is created, and what parts there are all the way down to the rectangle lists. How this list is translated into graphics onto a screen. XAA is fairly well documented. XVideo is a bit rough. XRender has somewhat real documentation, and you can read the thoughts of the designers on the public lists.
Let me know when the revolution starts..
Pan
I totally agree! Yeah man, we'll just ban evil!
Even Plato acknowledged a warrior class! It's simply not possible for a Utopia of this nature here. Whatever your moral arguments for or against Iraq, it is here and now. You want soldiers on your side in this situation.
Pan
Actually the 15% off $50 dollars gave me a 29$ discount. It meant 15% off purchase price up to $50 dollars.
Pan
Yes, there are programs that do this. One interesting thing is fixed floating point. It's basically where you say one integer register can hold enough significant digits to emulate a floating point in limited situations (games, some rendering types, etc)
The big advantage is that integer operations tend to get ganged through the ALU at a high rate (and on some cpu's simultaniously.. i.e. alpha).
This could also be useful for brute-force cracking on RSA, etc..
Anyhow.. YMMV, limited liability, and always seek legel advice before investing in risky ventures.
Pan
Just a couple of comments...
Direct3d's API works across all video platforms so long as the drivers are compliant. No extensions, so you write your code and it works on all cards. Nonsupported features are emulated in software.
Bit of a herring there.. OpenGL was the first HW assisted 3D graphics API to actually support software fallbacks. Any "compliant driver" can be designed to do the same, whether in OpenGL or D3D.
Direct3d is meant for what we have today instead of being the same set of commands which have been extended over and over again.
This is only through the wonders of COM.. Which also means a lot of your CPU time is spent doing COM translations. I would argue though that having a C++ or some object oriented API for OpenGL would be wonderous.
Pan
So ISP's are now going to be CA's as well? Or will Verisign wrap up the market with authentication DNS servers as well?
Pan
dd if=/dev/urandom of=/dev/hdc
Repeat as necessary.
Pan
Even Bjorn's paper says..
"Comparisons of languages, libraries, and programming styles are notoriously tricky, so please do not draw sweeping conclusions from these simple tests."...
Pan
Hrm.. I've built a system that does this using 4x redundancy, though for security products not trading. ;-)
We were able to scale 128 million transactions per hour on 16 PIII-500 boxes. The key to our transactions were shallow FIFO queues. Before transactions were executed, the queue list would require certain "sub transactions" to have been performed on 4 separate, though "unreliable" systems. Only 2 transactions were required for completion, and at that stage we executed the transaction and responded. Failed or timed-out transactions would be de-queued and responded to.
The key to speed was shallow (one level deep) message queues, and only guaranteeing 50% of the sub-transactions. Since each component was redundant, once 2 were completed we had full verification of the security operation. When the transaction commenced, we marked the transaction ID as "in use" and all other sub-transactions were tossed.
Another interesting tidbit on our system was the up-front design requirement for multiple data centers. So all the "sub-transactions" were actually performed on systems in CA, TX, IL, and VA. That's why they're "unreliable" (and latency-prone as well).
Also, store and forward transactions were distributed across the 4 locations, with the requirement that transactions could not be lost. Since our data was stored securely, we couldn't just simply keep a copy in a file or table, so we stored the Transaction ID's into a primary transaction engine (system which controlled top level interface to outside systems, somewhat like a transaction switch). When system load fell below certain percentages, the transaction engine would execute recommit commands to the data storage switches and build new keys for the data based on timestamps, and then recommit them back to all the data storage systems.
Our biggest performance hurdle was the fact that security algorithms are very slow to work with. The best way to cope with 2-3 second transactions was through massive parallelism of the transactions, so that even though all transactions were 2 or 3 seconds, we could perform a thousand per second per location.
Pan