In order to stop someone (or a country) that is willing to go to any means necessary, you have to be willing to go to any means necessary to stop them.
No. "By any means necessary" is not an option.
If they were stronger, they would have been right, because people today wouldn't know any different.
How can you say that? The Maquis clearly thought the Axis was wrong. German dissidents clearly thought the Axis was wrong. Both groups were clearly weaker than the Axis powers.
Most everyone today agrees the Crusades were wrong. Same goes for the Inquisition. It did not take the absolute defeat of the aggressor to come to this conclusion.
Causes are not right and wrong based on some higher level, it all depends on your point of view.
No. Grey areas exist, but they exist because of our limited understanding of the universe. This is why education, thought and prayer are so important.
Moral relativism makes it too easy to sidestep our responsibilities.
"Just War Theory?" A theory? I thought you were claiming that this was a moral absolute, not a theory.
I'm referring to the concept as is generally done when talking about the subject. It's a theory in that it is a concept developed by early Church fathers (through much prayer, reflection and argument) that falls far short of the Christian ideal of pacifism. It's a result of our flawed human nature that such a concept is needed.
Certainly grey areas exist, as I mentioned in a follow-up post.
In your orig. posting you said:
There is never any justification for the murder of innocents.
Actually, I did not write that. But it is still true. Murder implies that important element of intent. A just action can never intend to kill innocents.
yet now you say that there are, in fact, times when killing an innocent is acceptable. "If I
I wouldn't say "acceptable." "Evil byproduct" is perhaps a better description.
don't tareget them but they unfortunatly get hit, well that's ok because its war. Unfortunate, but
No, it's absolutely not ok. But it is true that sometimes actions must be taken to halt evil and those actions can cause harm. It is still something we need to reconcile within ourselves, to each other and to God. We still have to answer for our actions.
acceptable." So how does that jive with your statement that there is never any justification for murder of innocents? Maybe that statement isn't so absolute after all.
As I said, intent plays a large role.
Note that while individual acts can be unjust, an overall campaign can still be considered just.
Again, you've just contridicted your "murder is never justified" statement. Now, as long as its part of a just campaign its ok.
Nowhere did I say that. The mentioned actions were absolutely not ok. The overall goal of defeating the Axis and the umbrella action to accomplish this, however, were just.
Absolute means absolute. No excuses.
You're absolutely right.:)
If we as a country would more carefully consider the justness of our policies...
And there in lies the rub. Who defines what is just?
God does.
You defined a just war as a war with an explicitly stated and achievable goal. Are you telling me that's the ONLY criteria for "justness?"
Of course not! There's the obvious requirement that the goal itself be just.
The Nazi's had an explicitly stated and achieveable goal. Was thier cause just?
If good and evil aren't relative, then you'd think someone could come up with a better method of asserting that their personal views on the subject are correct than using force against those who disagree.
The problem is that most people don't carefully consider and reflect upon their choices enough. Are there grey areas? Of course there are and reasonable people can disagree. But disagreement != hatred. Some things are clearly wrong.
The usual cop-out is to simply claim that those who believe differently do so intentionally due to moral bankruptcy and "know the truth in their hearts."
Intentionally? I don't think I've heard that from any reasonable person (and no, the Christian Coalition is not reasonable:)). Through serious education, prayer and reflection people can agree on a wide variety of issues.
No, not even then. The Just War Theory states pretty explicitly that attacking civilians is a no-no. "Collateral damage" must be minimized. There is recognition that innocent people will likely be hurt or killed but they cannot be targets. This includes civilians who might "just happen" to get killed in an attack on a military objective in a densely poulated area. Such an attack would not be just.
A Just War is also a war with an explicitly stated and achievable goal. The current effort certainly lacks both.
Was the US nuking Japan "evil?"
Were the Allies bombing of Dresden "evil?"
Were the Germans bombing of London "evil?"
"Evil" is a loaded term, but yes, you could describe all of these as such. Note that while individual acts can be unjust, an overall campaign can still be considered just.
I guess it depends on which side you are on.
I don't think so. beagle is right. Good and evil are not relative.
Don't get me wrong. I think 9/11 was pretty damn "evil." But the key words there are "I think."
It's this moral relativism that continually gets us into trouble. If we as a country would more carefully consider the justness of our policies, there'd be a lot more peace in this world. Even so, this does not, not, not justify what was done on Sep. 11 and those responsible should be brought to justice.
It is IMHO much more effective to write your legislators directly. Send it snail mail so it gets more attention and is archived. Copy it electronically if you feel it needs to arrive in a timely fashion.
The petition is rather vague. Letter-writing allows you to be more specific in their views. You can mention the cryptography and wiretapping proposals specifically, along with your views on what the U.S. response should be.
senate.gov and house.gov have all the information you need. Writing to legislators who do not represent your area of the country is not very effective. Legislators tend to listen much more closely if you can kick them out of office.
Was not his victory in Missuori and Florida highly suspect?
No. It has pretty much been proven a valid result. Ever wonder what happened to all those recounts various papers were conducting? Guess they never found anything interesting to report!
Was it not plainly obvious that the SC ruled along party lines to support the questionable #'s from Fla.?
The Court often rules along party lines. That's why appointing judges is so important.
That Bush was elected is certainly debatable.
No, it is not.
It may be correct, but there is an equal chance that it isn't.
No, there isn't.
Surely you do not dismiss the numerous irregularities WRT Bush's rise to Power.
They're only irregular in that this particular election was extraordinary in its close vote.
The original poster and various followers-up are right. RMS discredits his own argument with such frivolous statements.
Odd thing was they were mostly continuing research started by others.
Um...that's what research is. Only super-humans like Einstein get to publish freestanding papers. Very little research is breathtakingly innovative. There's that whole "standing on the shoulders of giants" thing, you see.
Re:"Just throw more hardware at it"? Not always.
on
Software Aesthetics
·
· Score: 1
I see your point with respect to comments. However, unoptimized code and inefficient algorithms are unacceptable for many server-side tasks
Which is why I qualified my statement with "usually.":) For the most part, development costs refect the time spent coding, debugging and maintaining. Good design can cut down on this. The sorts of memory problems discussed above can usually be solved with good design. Good design doesn't require the programmer to micromanage.
Of course, any exponential-time algorithm will swamp any machine now or in the future given a sufficiently large dataset.:)
I've done projects in C, VB (im not proud), C++ (yep MFC et al, 5 years)
Please don't take MFC as a good example of C++ code!:)
However, I also like to think of efficiency in terms of developer hours, support and maintainence.
Absolutely. Programmer time is usually much more important than machine time.
It frees the programmer and maintainer of the details of memory management. Not that a developer doesn't have to understand memory management and the implications of holding references to objects, but that allocation and freeing of memory isn't a constant requirement.
I like the fact that you recognize all developers need to understand memory management. I've seen too many comments about how Java will save the world through garbage collection. They don't realize that you actually have to release the reference before it gets cleaned up.
In my C++ experience, holding onto references causes far more memory bloatedness than real leaks. People don't spend enough time profiling their code.
Every C++ program of sufficient complexity has to be tested and debugged for memory leaks - someone always forgets to add a destroy call somewhere.
Often this is due to poor code/library design. Judicious use of auto_ptr and the various smart pointer classes available can eliminate most of these problems. auto_ptr is particularly useful in conjunction with exception handling.
Constructors and destructors are wonderful things. Often a few well-designed classes can solve these sorts of problems. Now, one thing I will gripe about is the need to declare destructors virtual. That's bitten me badly in the past! Thankfully, good compilers warn about it.
I won't even mention buffer overflow problems
These should be a non-issue with C++. Anyone using bare arrays in C++ code should be shot.:) Of course, you can overflow std::vector but at least you won't overwrite the stack. But then I'd have to ask why you aren't using std::istringstream or an equivalent (possibly custom) streaming system.
People drill them for simple questions that have already been answered in the FAQ. Why do they deserve polite responses if they can't be bothered to look at the standard docs for the group?
Because newbies are called that for a reason. They don't know what a FAQ is, much less where to find it! Sure, ideally people would lurk on the group for a while (at least a week or month until the regular FAQ/HOWTO gets posted), but sometimes folks don't have that luxury.
I used to be polite to newbies, but after the first 50, i gave that up. Now, I either ignore them or else flame them if they're being dicks.
Flaming is not an option. We should treat electronic discussions just as we would treat a face-to-face discussion. Would any of us really flame someone in public, in front of thousands of people?
Civility is not really all that difficult to master and in the long run we'll all look better for it.
Oops, I found the phrase "single discussion archive" in my post.:)
What I'm talking about is the single (per domain) USENET feed that is kept on a server. This is much more efficient that having each user in the domain keep an archive.
Huh? Where did I mention a single discussion archive? Perhaps you misunderstood. I was suggesting that if individual groups want to archive their messages an e-mail gateway to an archive server (one per group) can be set up. Those servers' archives can be mirrored, of course!
You have a point but why get into the hassle of registering a USENET newsgroup?
I only mentioned USENET by name to provide some context. Clearly it needs an update. Many things like group creation are too difficult.
I think USENET is an outdated medium for discussions.
It's only outdated because we haven't kept it
up-to-date.
If you have a server, set up your slash-based discussion group, if not, just set up one of the free ones.
I find web-based discussions to be incredibly inconvenient. I like to be able to handle my mail and news from a single client with similar (often identical) interfaces. The latency encountered when reading posts on the web is very irritating to me. Posting is also very difficult. With news and mail I can use all the nice utilities on my machine for spellchecking, etc. I suppose these could be integrated into browsers but as yet they haven't been.
I mentioned the first two problems in my original post. Clearly USENET needs an update.
As for the third, I've not experienced this problem. Usually conversation ends because someone gets tired of it.:) But if this is a problem, then yes, something needs to be done.
Well, once you deny a group you can always set up an automatic reply mechanism. No reason you have to personally deal with every request. You can do the same thing for groups you already know won't be carried.
really? You call potentially botched message sending, inequal distribution and possible loss of messages due to news queue rollover good?
You've missed the point. USENET is a flawed system. I don't think many people will argue with that. But why aren't we fixing it? Or just start over with something new if USENET admins don't want to give up their power.
Granted, USENET has its share of problems. It's hard to create groups/run a server and the S/N ration can get very low. Perhaps a slightly tweaked version is needed.
Yeah they have something like that it's called SMTP
No. You're using a hammer as a wrench. Use e-mail for what it was designed for: personal and individual communication. Leave mass discussion to a system designed to handle it.
I mentioned rollover
e-mail has rollover problems as well. In fact they're worse because all the duplicate mails means rollover will happen that much faster. Keeping a single discussion archive should solve this problem. It's only because USENET is inconvenient to admin that it takes up so much disk space. I doubt even 50% of the groups umich carries are even read by anyone here.
not to mention the current lack of archieving in usenet right now for all but the most important groups.
Again, we could design a system to easily set up group archives. At the very least one could gateway a USENET discussion through e-mail to one single server that takes care of archiving. This is a stopgap solution at best, though.
My opinion: make it more open. Make more open portals and free methods to access it (ie. mailandnews.com which unfortunately is having a problem with it's newsfeet: any light?). And make sure they carry almost all the groups you can think of. And maybe have a nice public archive service at least for the text groups.
All good suggestions. Of course, the devil's in the details. It needs to be flexible enough to allow site admins to tailor it to their own needs (i.e. to not carry most groups, etc.). Archiving is better handled on an individual group basis rather than relying on a central server or two.
Why do most software projects (for example) use mailing lists rather than USENET?
Easy. It's too damn hard to set up a well-propagated newsgroup:
Exactly! USENET needs an overhaul. Why do we get new e-mail RFC's but none for USENET (or at least nothing well-publicized)? USENET should have the ability to easily create member-only groups.
Unfortunately, USENET also requires users to get their admins to carry the groups. This is a tough one to solve. Perhaps we could design things so that subscribing to a group not carried by a provider would automatically send a request to the admin for the group to be carried. The admin could then notify the user of any denials and state the reasons for such action.
The point is that there are good (unix) tools available to handle the flood of email; in particular Gnus alone can allow one to get it under control with very sophisticated sorting and scoring/filtering capabilities.
This is the same argument for dealing with SPAM in general. I don't buy it. These are solutions to the symptoms, not the problem. Yes, you can filter. But that doesn't stop the waste of bandwidth and disk space. There is absolutely no reason mailing lists should be needed. It's only because we've dropped the ball on USENET that things have progressed to the sorry state they're in now.
Spam is not the biggest problem on USENET -- it's people that have been there forever and think they deserve the right to rough up the newbies.
That is very true. The same thing happens on certain mailing lists. This has more to do with the people involved than the medium, though the medium does tend to bring out the worst in people. It's easy to dispense with civility when you don't have to address someone face-to-face.
As I see it, mailing lists are a big problem. They are trying to do something that is handled much better elsewhere: USENET. It makes much more sense to use something designed for mass distribution and discussion.
Granted, USENET has its share of problems. It's hard to create groups/run a server and the S/N ration can get very low. Perhaps a slightly tweaked version is needed.
Moderation helps but places a large burden on the moderator. Mailing lists (at least the ones I subscribe to) don't seem to have a SPAM problem nearly as large as on USENET. Perhaps because it's easy to set up a mailing list that requires registration to post. Why doesn't USENET have anything like that?
It's a complete waste of resources to have everyone in a domain store separate copies of discussion messages when one USENET archive could be available to all.
What's the trick? Why do most software projects (for example) use mailing lists rather than USENET? How can we take back USENET?
On the Alpha, a load into R0 acts as a memory prefetch. If the IA64 architecture defines this as faulting, separate prefetch instructions will be needed. Now, they might be needed anyway to get various levels of service/semantics, but a "basic" prefetch is often implemented this way.
Did you ever wonder why Mozilla is continually falling behind schedule? Because people like the Konqueror team decide to go off on their own instead of working for the good of the community.
Don't go blaming Konqueror or other browser projects for Mozilla's problems. I love Mozilla. I use it as my browser. But it has problems and those problems are there in spite of Konqueror, not because of it.
Mozilla was there first, and it deserves the support of the community.
Just because a project is "first" doesn't mean it should get a monopoly on developer resources. Should the same be said for KDE over GNOME?
If the free software community wants to make a good impression on the business world (and it may
This is a flawed assumption. The Free Software Community does not exist for the business world or any other world outside that of the people hacking code.
That said, if a project wants to court business, I'm all for it. But don't assume that is the goal of every project.
Here's an idea: before starting your new project, check to see if someone is already working on a similar project. Had the Konqueror team observed this little suggestion, the whole Konqueror fiasco
could have been avoided.
I'm sure the Konqueror team was well aware of Mozilla when they started. The projects have different architectures and different goals. Mozilla will never be the integrated browser Konqueror is. ioslaves are something Mozilla will not have any time in the near future.
There are lots of duplicate projects out there. I'm thinking of starting my own shortly. Why? Because all the other similar projects don't have the goals I have and their designs are clearly biased toward something I have no interest in. Furthermore, such designs are nearly impossible to "fix" properly to accomplish what I need. These are fundamental, core architectural decisions that can't be patched around. Better to start from scratch and build something the Right Way to accomplish my goals.
Not to mention the complete lack of documentation and code comments.:(
So, overall, I'm not sure where the 328 number comes from.
Probably 128 integer regs + 128 float regs + 64 branch/predicate regs (note: NOT general-purpose) + miscellaneous regs like the IP, etc.
While the P3/P4 have lots of registers, they aren't registers in the sense most people think about them. They solve the dynamic antidependecy problem. The static data allocation problem is a separate beast. Those renaming registers aren't visible to the compiler so you'll still have the same number of memory operations in the program.
Same deal with SMT/Hyperthreading. More registers are needed, but they aren't the sorts of registers the compiler can use.
It's interesting that a write to R0 is defined to fault. Is this just for Itanium or is it an IA64 architectural decision? If so, it seems like a very poor one to me.
Slashdot Mirror
No. "By any means necessary" is not an option.
How can you say that? The Maquis clearly thought the Axis was wrong. German dissidents clearly thought the Axis was wrong. Both groups were clearly weaker than the Axis powers.
Most everyone today agrees the Crusades were wrong. Same goes for the Inquisition. It did not take the absolute defeat of the aggressor to come to this conclusion.
No. Grey areas exist, but they exist because of our limited understanding of the universe. This is why education, thought and prayer are so important.
Moral relativism makes it too easy to sidestep our responsibilities.
I'm referring to the concept as is generally done when talking about the subject. It's a theory in that it is a concept developed by early Church fathers (through much prayer, reflection and argument) that falls far short of the Christian ideal of pacifism. It's a result of our flawed human nature that such a concept is needed.
Certainly grey areas exist, as I mentioned in a follow-up post.
Actually, I did not write that. But it is still true. Murder implies that important element of intent. A just action can never intend to kill innocents.
I wouldn't say "acceptable." "Evil byproduct" is perhaps a better description.
No, it's absolutely not ok. But it is true that sometimes actions must be taken to halt evil and those actions can cause harm. It is still something we need to reconcile within ourselves, to each other and to God. We still have to answer for our actions.
As I said, intent plays a large role.
Nowhere did I say that. The mentioned actions were absolutely not ok. The overall goal of defeating the Axis and the umbrella action to accomplish this, however, were just.
You're absolutely right. :)
God does.
Of course not! There's the obvious requirement that the goal itself be just.
Clearly not.
The problem is that most people don't carefully consider and reflect upon their choices enough. Are there grey areas? Of course there are and reasonable people can disagree. But disagreement != hatred. Some things are clearly wrong.
Intentionally? I don't think I've heard that from any reasonable person (and no, the Christian Coalition is not reasonable :)). Through serious education, prayer and reflection people can agree on a wide variety of issues.
No, not even then. The Just War Theory states pretty explicitly that attacking civilians is a no-no. "Collateral damage" must be minimized. There is recognition that innocent people will likely be hurt or killed but they cannot be targets. This includes civilians who might "just happen" to get killed in an attack on a military objective in a densely poulated area. Such an attack would not be just.
A Just War is also a war with an explicitly stated and achievable goal. The current effort certainly lacks both.
"Evil" is a loaded term, but yes, you could describe all of these as such. Note that while individual acts can be unjust, an overall campaign can still be considered just.
I don't think so. beagle is right. Good and evil are not relative.
It's this moral relativism that continually gets us into trouble. If we as a country would more carefully consider the justness of our policies, there'd be a lot more peace in this world. Even so, this does not, not, not justify what was done on Sep. 11 and those responsible should be brought to justice.
The petition is rather vague. Letter-writing allows you to be more specific in their views. You can mention the cryptography and wiretapping proposals specifically, along with your views on what the U.S. response should be.
senate.gov and house.gov have all the information you need. Writing to legislators who do not represent your area of the country is not very effective. Legislators tend to listen much more closely if you can kick them out of office.
Yes, but that is completely irrelevant.
No. It has pretty much been proven a valid result. Ever wonder what happened to all those recounts various papers were conducting? Guess they never found anything interesting to report!
The Court often rules along party lines. That's why appointing judges is so important.
No, it is not.
No, there isn't.
They're only irregular in that this particular election was extraordinary in its close vote.
The original poster and various followers-up are right. RMS discredits his own argument with such frivolous statements.
Um...that's what research is. Only super-humans like Einstein get to publish freestanding papers. Very little research is breathtakingly innovative. There's that whole "standing on the shoulders of giants" thing, you see.
Which is why I qualified my statement with "usually." :) For the most part, development costs refect the time spent coding, debugging and maintaining. Good design can cut down on this. The sorts of memory problems discussed above can usually be solved with good design. Good design doesn't require the programmer to micromanage.
Of course, any exponential-time algorithm will swamp any machine now or in the future given a sufficiently large dataset. :)
Please don't take MFC as a good example of C++ code! :)
Absolutely. Programmer time is usually much more important than machine time.
I like the fact that you recognize all developers need to understand memory management. I've seen too many comments about how Java will save the world through garbage collection. They don't realize that you actually have to release the reference before it gets cleaned up.
In my C++ experience, holding onto references causes far more memory bloatedness than real leaks. People don't spend enough time profiling their code.
Often this is due to poor code/library design. Judicious use of auto_ptr and the various smart pointer classes available can eliminate most of these problems. auto_ptr is particularly useful in conjunction with exception handling. Constructors and destructors are wonderful things. Often a few well-designed classes can solve these sorts of problems. Now, one thing I will gripe about is the need to declare destructors virtual. That's bitten me badly in the past! Thankfully, good compilers warn about it.
These should be a non-issue with C++. Anyone using bare arrays in C++ code should be shot. :) Of course, you can overflow std::vector but at least you won't overwrite the stack. But then I'd have to ask why you aren't using std::istringstream or an equivalent (possibly custom) streaming system.
Because newbies are called that for a reason. They don't know what a FAQ is, much less where to find it! Sure, ideally people would lurk on the group for a while (at least a week or month until the regular FAQ/HOWTO gets posted), but sometimes folks don't have that luxury.
Flaming is not an option. We should treat electronic discussions just as we would treat a face-to-face discussion. Would any of us really flame someone in public, in front of thousands of people?
Civility is not really all that difficult to master and in the long run we'll all look better for it.
What I'm talking about is the single (per domain) USENET feed that is kept on a server. This is much more efficient that having each user in the domain keep an archive.
Huh? Where did I mention a single discussion archive? Perhaps you misunderstood. I was suggesting that if individual groups want to archive their messages an e-mail gateway to an archive server (one per group) can be set up. Those servers' archives can be mirrored, of course!
I only mentioned USENET by name to provide some context. Clearly it needs an update. Many things like group creation are too difficult.
It's only outdated because we haven't kept it up-to-date.
I find web-based discussions to be incredibly inconvenient. I like to be able to handle my mail and news from a single client with similar (often identical) interfaces. The latency encountered when reading posts on the web is very irritating to me. Posting is also very difficult. With news and mail I can use all the nice utilities on my machine for spellchecking, etc. I suppose these could be integrated into browsers but as yet they haven't been.
As for the third, I've not experienced this problem. Usually conversation ends because someone gets tired of it. :) But if this is a problem, then yes, something needs to be done.
Well, once you deny a group you can always set up an automatic reply mechanism. No reason you have to personally deal with every request. You can do the same thing for groups you already know won't be carried.
You've missed the point. USENET is a flawed system. I don't think many people will argue with that. But why aren't we fixing it? Or just start over with something new if USENET admins don't want to give up their power.
No. You're using a hammer as a wrench. Use e-mail for what it was designed for: personal and individual communication. Leave mass discussion to a system designed to handle it.
e-mail has rollover problems as well. In fact they're worse because all the duplicate mails means rollover will happen that much faster. Keeping a single discussion archive should solve this problem. It's only because USENET is inconvenient to admin that it takes up so much disk space. I doubt even 50% of the groups umich carries are even read by anyone here.
Again, we could design a system to easily set up group archives. At the very least one could gateway a USENET discussion through e-mail to one single server that takes care of archiving. This is a stopgap solution at best, though.
All good suggestions. Of course, the devil's in the details. It needs to be flexible enough to allow site admins to tailor it to their own needs (i.e. to not carry most groups, etc.). Archiving is better handled on an individual group basis rather than relying on a central server or two.
Exactly! USENET needs an overhaul. Why do we get new e-mail RFC's but none for USENET (or at least nothing well-publicized)? USENET should have the ability to easily create member-only groups.
Unfortunately, USENET also requires users to get their admins to carry the groups. This is a tough one to solve. Perhaps we could design things so that subscribing to a group not carried by a provider would automatically send a request to the admin for the group to be carried. The admin could then notify the user of any denials and state the reasons for such action.
This is the same argument for dealing with SPAM in general. I don't buy it. These are solutions to the symptoms, not the problem. Yes, you can filter. But that doesn't stop the waste of bandwidth and disk space. There is absolutely no reason mailing lists should be needed. It's only because we've dropped the ball on USENET that things have progressed to the sorry state they're in now.
That is very true. The same thing happens on certain mailing lists. This has more to do with the people involved than the medium, though the medium does tend to bring out the worst in people. It's easy to dispense with civility when you don't have to address someone face-to-face.
Granted, USENET has its share of problems. It's hard to create groups/run a server and the S/N ration can get very low. Perhaps a slightly tweaked version is needed.
Moderation helps but places a large burden on the moderator. Mailing lists (at least the ones I subscribe to) don't seem to have a SPAM problem nearly as large as on USENET. Perhaps because it's easy to set up a mailing list that requires registration to post. Why doesn't USENET have anything like that?
It's a complete waste of resources to have everyone in a domain store separate copies of discussion messages when one USENET archive could be available to all.
What's the trick? Why do most software projects (for example) use mailing lists rather than USENET? How can we take back USENET?
Yup, I fudged my history a bit. :) Apologies to the kfm/Konqueror team.
Of course, this doesn't invalidate my point.
On the Alpha, a load into R0 acts as a memory prefetch. If the IA64 architecture defines this as faulting, separate prefetch instructions will be needed. Now, they might be needed anyway to get various levels of service/semantics, but a "basic" prefetch is often implemented this way.
Don't go blaming Konqueror or other browser projects for Mozilla's problems. I love Mozilla. I use it as my browser. But it has problems and those problems are there in spite of Konqueror, not because of it.
Just because a project is "first" doesn't mean it should get a monopoly on developer resources. Should the same be said for KDE over GNOME?
This is a flawed assumption. The Free Software Community does not exist for the business world or any other world outside that of the people hacking code.
That said, if a project wants to court business, I'm all for it. But don't assume that is the goal of every project.
I'm sure the Konqueror team was well aware of Mozilla when they started. The projects have different architectures and different goals. Mozilla will never be the integrated browser Konqueror is. ioslaves are something Mozilla will not have any time in the near future.
There are lots of duplicate projects out there. I'm thinking of starting my own shortly. Why? Because all the other similar projects don't have the goals I have and their designs are clearly biased toward something I have no interest in. Furthermore, such designs are nearly impossible to "fix" properly to accomplish what I need. These are fundamental, core architectural decisions that can't be patched around. Better to start from scratch and build something the Right Way to accomplish my goals.
Not to mention the complete lack of documentation and code comments. :(
Probably 128 integer regs + 128 float regs + 64 branch/predicate regs (note: NOT general-purpose) + miscellaneous regs like the IP, etc.
While the P3/P4 have lots of registers, they aren't registers in the sense most people think about them. They solve the dynamic antidependecy problem. The static data allocation problem is a separate beast. Those renaming registers aren't visible to the compiler so you'll still have the same number of memory operations in the program.
Same deal with SMT/Hyperthreading. More registers are needed, but they aren't the sorts of registers the compiler can use.
It's interesting that a write to R0 is defined to fault. Is this just for Itanium or is it an IA64 architectural decision? If so, it seems like a very poor one to me.
No too difficult, really. :)