The "fine technical details" should pale next to common sense, and often the best sanity check is to have people who know nothing about the subject hear about who has done what.
?Coyote? -> Coyotos [Jonathan Shapiro's capability-based kernel]. There was talk about a move to either Coyotos or L4sec, something about L4 not doing protected IPC? Coyotos is a long way from being finished itself, and even then it is nothing like Mach so all the servers will probably need to be ported:P.
Coyotos is pretty revolutionary- but it's moving very, very slowly. We need more developers, cap'n!
Thanks, I was going to say something similar. The Hurd doesn't have developers coming out of its ears like Linux does, but if you want to run it on your x86 machine, Debian have a distribution of it that works today.
The graphics libraries are quite an important part of a mac application, and some things that an application can assume to be there under OS X have no direct parallel in X. link. I don't think an API wrapper from Quartz to X etc would be trivial:P
I wouldn't say that: Itanium is a nice idea [even compared to RISC], it means you can cut the fat from things like Out of Order Execution. However, the floating point units on the Itanic don't appear to be up to Alpha quality.
If consumers want to be locked in to crappy closed source software, let them be. The best way to bring people to Free software is to let them try the alternative.
Simple capability management? I expect we'll get there eventually. Most of the microkernels in development today have this functionality built in. On the other hand, they also have the possibly-vaporware feature built in;) Expect to live with VMs until Linux goes the way of the dinosaur, in 64 million years.
Re:A (Fruit) Fly in the Ointment?
on
GCC 4.2.1 Released
·
· Score: 2, Insightful
I think they mean will it stop them contributing, not using.. I'd say unlikely. If Apple, say, were likely to put their own patented code into gcc before, I don't see what the big difference is now- anyone could have taken their patented code before, straight from gcc, legal or not. Apple mainly contribute to the Objective C portions, and I don't think any of it would be patented. I'm not sure what IBM contribute, but given the recent patent deals, I doubt it would be a problem if they did contribute patented stuff.
The data copy can probably be taken care of through memory remapping or DMA*, but you're right, that context switch time is still completely silly. One wonders how Intel could have stuck with such brain-dead behavior for so long, and why they aren't going to do anything about it. Furthermore, I don't see how zeroing a cache can take so long! That's nearly as slow as moving zero into each slot by hand!
*Something mk designers frequently overlook. I don't think there's anything particularly bad about address spaces that overlap. I think this is the middle ground between microkernels and monolithic kernels that is The Right Thing- address space separation except where it's obvious that it's a good thing. Sometimes it's good to pass pointers to entire data structures around, and not particularly difficult on x86.
I can do that here with the scroll button, but otherwise it's not something that has ever happened to me, on Windows or Linux.
The one thing I would most like to see in Firefox would be threading [or even separate processes for different s/tlds and windows] for tabs and the GUI. It really is a pain having to wait until a tab has started before scrolling down or clicking another link. With that kind of model, I bet there wouldn't be much difficulty adding a compile time option for no tabs.
Come to think of it- I bet you can do it! Go on, get hacking!:)
What are you saying, the down arrow is too difficult now? Does the effort of using the down arrow severely outweigh that of setting up a bookmark keyword? This is nothing but pedantry- the main thing is, the history in the URL line is just as usable, and I'd wager it's not fat at all.
Noscript blocks Javascript by default and allows you to turn it on either temporarily or permanently for different sites. It's the one plug-in I'd take to a desert island. Or something like that:P
You know, the cache is a big part of it. The main reason Firefox is so big is because it assumes that the operating system is better at paging out cache than itself [which is usually correct, though it gets a lot of flack for this]. I was going to say that this doesn't explain the large size at startup, but then again, it's not much larger than the latest Windows Dead Messenger.
But then I have to ask, what has this got to do with firefox? FF hardly is the reason for the proliferation of flash on the web, and if you don't like it you don't have to download it. It's not part of firefox at all.
you do realise that 'slash' will bring up slashdot even without it being in bookmarks, if it's in your history? I never even bother setting up a bookmark for/., rather just type 's l down-arrow'.
You are trolling, right? Firefox doesn't open any tabs unless you tell it to, by default, and the history list needs a Ctrl-H or Alt-S to be shown [although I gather you mean forms history].
Although your post is cute, you actually have a point. You can argue about how changes to the kernel affect performance, but otherwise they aren't likely to make a difference to the GNU/Linux experience. Maybe we should be actively pointing our developer friends away from romantic dreams of kernel hacking and on to more overlooked projects. Or even wine..
That's what, 80,000 cycles plus? That's ridiculous. All to save some registers and load a new LDT?
I agree on the Linux never becoming a microkernel, that would be a really ugly hack. But I think the kernels being developed today, the ones with memory essentially being a first class, sharable resource [such as Coyotos, which IIRC has an interface for sharing address spaces with other processes]. Once shared memory becomes simple and usable, we might see some nice new mk on top.
In a statement Thursday, Jim Cicconi, AT&&T's senior executive vice president for external and legislative affairs, said Mr. Martin's proposal was an "interesting and creative balance" that would not change the business models of AT&T and others. He said consumers would now have to "put up or shut up."
If the server has enough bandwidth to let me sit there uploading my file and tell me it's done, and enough bandwidth for the receiver to view their fat javascripty Hotmail page, they should probably have enough bandwidth to let receivers download the attachment, or at least notify users that it's on the way or has failed.
The coupons thing will probably bow to common sense in the end. As much as I'd like MS to be bitten by that, I'm quite sure it will get thrown out in any reasonable court of law. On the other hand, the premise of the deal will probably be enough to bite them hard in court, because any way you look at it they seem to be modifying and distributing GPL code without allowing users to redistribute it, or making subsequent users pay.
Then again, it's not like common sense prevails often when M$ are involved, so who knows what could happen?:)
It appears to be. The normal mathematical definition for the Mobius strip, IIRC, is a square with a pair of equivalence relations on the boundaries, but that's not a differential equation. I was actually under the impression that there was proof that the mobius strip could not be written in the way they speak of [I think they are talking about a Riemann manifold? Stress density, really now..] anyway, without access to the article, there's no way to know.
No, you can measure a subset of usage this way. Sites that call home often are going to get closed or disused very quickly. More and more people tend to whitelist active content, because it is increasingly annoying. As that gets worse, you can still use cookies in some cases, but that also assumes you refresh the site. This is a great way to measure usage of enterprisey AJAX pages, and maybe forums [although there's no advantage with this method here], but otherwise, it's not always going to work.
Slashdot Mirror
The "fine technical details" should pale next to common sense, and often the best sanity check is to have people who know nothing about the subject hear about who has done what.
?Coyote? -> Coyotos [Jonathan Shapiro's capability-based kernel]. There was talk about a move to either Coyotos or L4sec, something about L4 not doing protected IPC? Coyotos is a long way from being finished itself, and even then it is nothing like Mach so all the servers will probably need to be ported :P.
Coyotos is pretty revolutionary- but it's moving very, very slowly. We need more developers, cap'n!
Thanks, I was going to say something similar. The Hurd doesn't have developers coming out of its ears like Linux does, but if you want to run it on your x86 machine, Debian have a distribution of it that works today.
The graphics libraries are quite an important part of a mac application, and some things that an application can assume to be there under OS X have no direct parallel in X. link. I don't think an API wrapper from Quartz to X etc would be trivial :P
Isn't the Itanium essentially a repackaged Alpha?
I wouldn't say that: Itanium is a nice idea [even compared to RISC], it means you can cut the fat from things like Out of Order Execution. However, the floating point units on the Itanic don't appear to be up to Alpha quality.
If consumers want to be locked in to crappy closed source software, let them be. The best way to bring people to Free software is to let them try the alternative.
Simple capability management? I expect we'll get there eventually. Most of the microkernels in development today have this functionality built in. On the other hand, they also have the possibly-vaporware feature built in ;) Expect to live with VMs until Linux goes the way of the dinosaur, in 64 million years.
I think they mean will it stop them contributing, not using.. I'd say unlikely. If Apple, say, were likely to put their own patented code into gcc before, I don't see what the big difference is now- anyone could have taken their patented code before, straight from gcc, legal or not. Apple mainly contribute to the Objective C portions, and I don't think any of it would be patented. I'm not sure what IBM contribute, but given the recent patent deals, I doubt it would be a problem if they did contribute patented stuff.
The data copy can probably be taken care of through memory remapping or DMA*, but you're right, that context switch time is still completely silly. One wonders how Intel could have stuck with such brain-dead behavior for so long, and why they aren't going to do anything about it. Furthermore, I don't see how zeroing a cache can take so long! That's nearly as slow as moving zero into each slot by hand!
*Something mk designers frequently overlook. I don't think there's anything particularly bad about address spaces that overlap. I think this is the middle ground between microkernels and monolithic kernels that is The Right Thing- address space separation except where it's obvious that it's a good thing. Sometimes it's good to pass pointers to entire data structures around, and not particularly difficult on x86.
I can do that here with the scroll button, but otherwise it's not something that has ever happened to me, on Windows or Linux.
:)
The one thing I would most like to see in Firefox would be threading [or even separate processes for different s/tlds and windows] for tabs and the GUI. It really is a pain having to wait until a tab has started before scrolling down or clicking another link. With that kind of model, I bet there wouldn't be much difficulty adding a compile time option for no tabs.
Come to think of it- I bet you can do it! Go on, get hacking!
And that's something no one wants to see!
What are you saying, the down arrow is too difficult now? Does the effort of using the down arrow severely outweigh that of setting up a bookmark keyword? This is nothing but pedantry- the main thing is, the history in the URL line is just as usable, and I'd wager it's not fat at all.
Noscript blocks Javascript by default and allows you to turn it on either temporarily or permanently for different sites. It's the one plug-in I'd take to a desert island. Or something like that :P
You know, the cache is a big part of it. The main reason Firefox is so big is because it assumes that the operating system is better at paging out cache than itself [which is usually correct, though it gets a lot of flack for this]. I was going to say that this doesn't explain the large size at startup, but then again, it's not much larger than the latest Windows Dead Messenger.
Oh, of course. Gratuitous use of flash [and java/javascript] is probably the bane of anyone who wants to do something useful on the web. Just the other day I was trying to download the catalogue from this site. As you can see, there's an area entitled '2006 catalogue (click here)' that does absolutely nothing. The site might work if the person who wrote it spent a little less time writing the javascript to run it [view the source and hold on to your hat] and more time, you know, making something usable. Most of the time, flat HTML+CSS really is the best way to go, but that probably doesn't sit too well with many PHBs and ponytails.
But then I have to ask, what has this got to do with firefox? FF hardly is the reason for the proliferation of flash on the web, and if you don't like it you don't have to download it. It's not part of firefox at all.
you do realise that 'slash' will bring up slashdot even without it being in bookmarks, if it's in your history? I never even bother setting up a bookmark for /., rather just type 's l down-arrow'.
You are trolling, right? Firefox doesn't open any tabs unless you tell it to, by default, and the history list needs a Ctrl-H or Alt-S to be shown [although I gather you mean forms history].
I highly doubt "tabs, ..favourites, history lists" are the memory burner. Would be an interesting area to analyse, though.
Although your post is cute, you actually have a point. You can argue about how changes to the kernel affect performance, but otherwise they aren't likely to make a difference to the GNU/Linux experience. Maybe we should be actively pointing our developer friends away from romantic dreams of kernel hacking and on to more overlooked projects. Or even wine..
That's what, 80,000 cycles plus? That's ridiculous. All to save some registers and load a new LDT?
I agree on the Linux never becoming a microkernel, that would be a really ugly hack. But I think the kernels being developed today, the ones with memory essentially being a first class, sharable resource [such as Coyotos, which IIRC has an interface for sharing address spaces with other processes]. Once shared memory becomes simple and usable, we might see some nice new mk on top.
In a statement Thursday, Jim Cicconi, AT&&T's senior executive vice president for external and legislative affairs, said Mr. Martin's proposal was an "interesting and creative balance" that would not change the business models of AT&T and others. He said consumers would now have to "put up or shut up."
If the server has enough bandwidth to let me sit there uploading my file and tell me it's done, and enough bandwidth for the receiver to view their fat javascripty Hotmail page, they should probably have enough bandwidth to let receivers download the attachment, or at least notify users that it's on the way or has failed.
The coupons thing will probably bow to common sense in the end. As much as I'd like MS to be bitten by that, I'm quite sure it will get thrown out in any reasonable court of law. On the other hand, the premise of the deal will probably be enough to bite them hard in court, because any way you look at it they seem to be modifying and distributing GPL code without allowing users to redistribute it, or making subsequent users pay.
:)
Then again, it's not like common sense prevails often when M$ are involved, so who knows what could happen?
[IANAL]
It appears to be. The normal mathematical definition for the Mobius strip, IIRC, is a square with a pair of equivalence relations on the boundaries, but that's not a differential equation. I was actually under the impression that there was proof that the mobius strip could not be written in the way they speak of [I think they are talking about a Riemann manifold? Stress density, really now..] anyway, without access to the article, there's no way to know.
Well, no. In 2002 maybe [ESR wrote a great essay on how bad it was]. But it's not like that at all now.
No, you can measure a subset of usage this way. Sites that call home often are going to get closed or disused very quickly. More and more people tend to whitelist active content, because it is increasingly annoying. As that gets worse, you can still use cookies in some cases, but that also assumes you refresh the site. This is a great way to measure usage of enterprisey AJAX pages, and maybe forums [although there's no advantage with this method here], but otherwise, it's not always going to work.