If you have to sign something that says ``I did such-and-such'' and you did, you should sign it. If you have to sign something that says ``this is an origional idea, MyCompany is extremely innovative and nothing of this sort has been attempted before by anyone else, and it is necessary for us to have a patent for the advancement of this technology'' then if you don't agree with it, you shouldn't sign it.
It is all based on wording. If you agree with some of it, and not with others, cross it out, initial, and then sign the modified document.
Or, add stuff in, initial, and sign.
It's often pretty easy to get an app to use your own custom catalog (just alter your config file), which contains strings that can
exploit the problem.
Right, but even if you get an application to read your own catalog, it's still running as your userid. So it can only do as much damage as you could do normally.
Sure, you can change root-level programs to read from arbitrary files, but if you can change root-owned files then you can do anything you want in the system without having to make programs that do wierd things with buffer overflows and such.
that Napster is not stealing IP... it is merely a technology that allows people to exchange music stored in the MP3 format. Now, many people choose to distribute music that is not legal to exchange (typically). That's not napster's fault in the same way it's not the fault of the post office when someone sends a mail bomb.
Some would go farther to say that record labels are evil and the actual artists don't get any money anyway, blah blah blah, and with dissertations it's actually people stealing from the little guy, not the evil corporation. I don't know if that argument is valid or not, but just because napster (or gnutella, or whatever) allows people to distribute illegally copied materials doesn't make the technology itself illegal.
For the last time (I hope): Microsoft is NOT being taken to court because it is a monopoly. It's being taken to court because it used it's market power to crush competition and do other anti-competitive things.
It's not illegal (or wrong) to control 100% of the market because you have a good (or popular) product. It is illegal (and wrong) to use your large market share to kill companies you don't like.
Netscape, through their 4.x browser series, was free for personal use, but closed-source. Netscape did quite well in the 1.x series, not so well in the 2.x series, fairly well in the 3.x series without javascript or java, and then in the 4.x series it's not been so great. Anyway, Netscape decided that they needed to open up the source for Netscape so they could compete with the behemoth and for other reasons as well. So, they released the source to Netscape, and the mozilla comnunity (consisting of independant people as well as Netscape employees) started to tinker with it.
They found that it was bloated, poorly written, and poorly engineered. So they threw it all away and started over almost completely. Architecturally, the new Mozilla is much much better, and should be able to be more flexible over time than the old code tree. But it's taken a long time to get there. And it's not done yet. But it has many, many nice features.
Red Hat's Mission Statement / Corp Policy or whatever is that (from http://redhat.com/about) ``Red Hat shares all of its software innovations freely with the open source community under the GNU General Public License (GPL).'' Indeed, they tend to not talk about themselves in terms of a linux distribution company, but a company doing R&D, Integration, and distribution of Open Source software in general. Hence, the merger with cygnus.
The problem is with machines that block ICMP. Then you can't do MTU path discovery. This was the case with lots of ``secure'' sites, and when my ISP installed a transparent web proxy/cache they disabled ICMP on that too, meaning that every external web page stopped working.
Pump more RAM in your system. Bandwith has never increased as fast as processing speed or storage size. Thus, instead of reading to and from RAM every time, you stick a bunch of cache in your processor. Poof, with a little L1 and L2 cache you can run most of your processor out of cache (for many things, cache hits on the processor are like 99.9%). This is good, because your RAM has lots of latency in addition to not transmitting the information back to your processor very fast. Usually, going to RAM inflicts any where from 5 to 50 CPU cycles waiting for the RAM (or more!).
Disk is worse. While RAM is, say, 10ns, a disk access is 10ms. That's 1 Million times slower. But we do the same thing... we can keep off the disk if we keep everything in RAM. And since we are multitasking and the disk is sooooo slow, if a process reads from disk we just take it off the processor and wait until the disk comes back with information before we will allow the process to come back on to the processor.
Now, since we're really smart, we do clever things to make sure that we have to go to Ram (or disk) as few times as possible. In cache, we load more than just the one byte or word that we want, we load the words next to it into cache as well, as that means that the next read from memory will (hopefully) be in cache already. This increases the penalty for going to RAM, but tends to pay off. Likewise, when we read a byte from disk, we load the 32K after it (or whatever we decide is appropriate) so that if we have to read the next byte it's in memory already.
Thus, we have really decreased the problems with the bottlenecks we do have. And it's very important... half the pins on modern processors are ground or power. That doesn't leave too many pins for I/O, and if you want to increase bandwith you usually want to increase the width of the bus in addition to bus speed. But unless you're Cray, it's really hard to run 1024 signals into your processors.
That's not to say that faster disks don't help, but you're not relying on your disk's speed every time you read or write a byte to your disk (thank goodness). Instead of trying to get the fastest disk, you may just want to increase the disk cache (IE, get more RAM). On my workstation, I can run Netscape and my terminals and text editor, and after I have opened them once or so, they just sit in RAM. When Netscape crashes and I have to start it up again, it pulls almost all of the program out of memory instead of having to go to disk. Instead of spending the money on a 10,000RPM Ultra Wide Fast Loose SCSI card and drive, I bought another 128M of RAM, and for most desktopish stuff (and a lot of heavier I/O, even) I can beat someone with less memory but a bigger disk.
And on the MTU issue, why do you care how big a network packet you send? How does this help 99.9% of users? I think you were just looking for some big sounding acronym to toss arround.
Not at all. This was a big problem for us... we have a firewall that talks to our DSL line... but the DSL line has a MTU of less than 1500, which is the standard for ethernet. Which meant that all the machines, including the Mac, stopped working when trying to go outside the firewall. The only tweaker program we could find that would change the setting cost $30. For a setting which should be (and is for every other OS we use) available in the OS's settings.
And that third party software you talk about simply adjusts a setting in OpenTransport, which is part of the OS, thus the tweeker prorgam is freeware....
Nope, not freeware. Had a free 30 day demo.
The MTU is probably my biggest example of why I hate Windows and hate Mac more. On all the unix boxes, you type ``ifconfig eth0 mtu 1400'' or you can set the MSS on the default route (either fix the problem as they both do about the same thing). Under windows, you have to make a registry entry to fix the problem. With MacOS, you have to hunt your way around extremely little documentation and finally after reading mailing lists you find some random guy selling his little shareware tweaker.
That just seems like the MacOS way. Tons of annoying shareware for things that should just work and be in the distribution.
And FUD = "Fear Uncertainty and Doubt". How is Apple spreading any of this? I watched the keynote, and Steve Jobs was very clear about the dual G4's being the fastest PERSONAL computers out there for using Photoshop
Which is different from what he was saying before. I've seen and heard Jobs spread lies about how superior the Mac was to anything. I've seen Apple tell everyone that they will be making their OS modern for years. I've seen them convince their customers that if they have to switch to Windows that they'll be dealing with IRQ conflicts and editing their config.sys all day long. How many times have you seen Mac people, even here on slashdot, use terms like ``plug-and-pray'' that they're convinced the Intel architecture is plagued with. I don't use windows much, but the last time I added a new video card to a Windows machine it worked just fine. And it's certainly been years since I dealt with an IRQ conflict. RedHat is amazing at figuring what hardware is added or removed from a system and configuring it.
Not to mention most of the non-apple-done benchmarks I've seen put the G4s at a significantly lower performance level than even middle-of-the-road x86 chips. Try here, can't find some of the better comparisons I've seen.
If what you really meant was "Don't support Linux" then you are correct, but there is no ecomomic reason for them to do so at this time.
Apple supports their own OS pretty well. Apple supports Windows in such a way as to make it annoying to use, as if they are penalizing them for choosing another OS. And while they say that many of the QT codecs are cross-platform, they really mean that they are available on Mac and Windows using the QuickTime Player from Apple Only. That's not a cross-platform standard in my opinion.
What about 64 bit PCI (standard, backward compatible to the regular 32 bit you find in most PC's), Gigabit Ethernet (backward compatable to 10/100), AGP, PC100 RAM (granted they use th 3-2-2 or better only), USB periferals, etc do you find propritary.
Well, let's see. Why does Apple have these things? Is it because they're working towards compatability, or because they were having trouble getting hardware vendors to make specialized NuBus, LCBus, etc. cards when it only affected 10% of the potential market? Meanwhile, how much have they really done to make Macs interoperate well with PCs? Has apple put a SMB client in their OS yet? No! How about an NFS client? No! The only thing it ships with is Appletalk.
So I must say that Apple did the right thing in getting a clue and going for an OS that finally has pre-emptive multitasking, real multiuser support, etc. And they finally figured out that they couldn't do it without going to a completely new codebase. I remember back about 7 years ago Apple talking about their new OS with features that should be in a modern OS, and we've only seen it now.
But Apple trying to converge with other OS's? That'll be the day. I think that Apple, as a corporation, is probably equally as evil as Microsoft, it's just that they haven't been in a position to do too much evil. Apple's not evil you say?
Get other vendors to come up with clones, then refuse to license the OS after they finally get up and running.
Support QuickTime under windows, but do it in such a way as to make the software as annoying as possible. Put your big logo up at the front every time it's launched. Make your own wierd control panel. Add yourself to the desktop. Claim to want to play everything, but not work on half the stuff. And the only reason they do it at all is because if they didn't support windows nobody would use QT at all.
Don't support QT on anything but Windows
Ship horrible mice and keyboards with your products, and then go with the line ``you can replace them with something else if you want!'' instead of shipping a reasonable product.
And, perhaps most importantly,
All the FUD and lies they spread. Steve Jobs claiming that you can't buy a faster computer than a mac with the G4 -- besides the fact that you could get an UltraSparc or Alpha, many PII's at the time beat it out in most things. Telling people that MacOS was technologically superior to everything. Telling them that it was compatable and flexible. You can't even change the MTU without third-party software.
Don't get me wrong. MacOS is very easy-to-learn. They put some interesting stuff in their boxes. But I wouldn't trust the company. I especially wouldn't trust the company to converge with anything. I'll be watching Apple try to stay as much in ultra-proprietary-land as they can.
Motorola designed the HC11 specifically for automobile applications.
If you're sick of the assembly then get a different assembler... I seem to remember the HC16 assembly (which is much like HC11... I was writing rocket control softwarE) that I did was rather nice. Not quite as nice as PIC assembly, but...
NFS caching is mainly for reading. It's most helpful when I have large amounts of programs served by NFS, and want to cut down on network traffic.
And, I mean, we currently have NFS caching, but only for memory. You can't assign 2G of hard disk space for caching stuff read off of NFS.
The way read caching keeps consistency is using leases on the data... the server will contact the client if the data changes before the lease is up, and if the server can't reach the other client then the client is not assured correct data after the lease expires.
There are already too many responses to this article for this to ever get read, but....
What we really need is Auto-Make in our operating system.
What I mean is, I should never have to generate a postscript file. Or a DVI file. If I want to print out or view the document I just edited, the OS should know that it needs to look for the TeX file and generate the DVI, then generate the postscript. It should cache them and delete them if I need the space.
Also, the operating system should generate binaries automatically from source. It should profile the binaries using spare CPU cycles and when the system is idle re-compile to try to optimize the parts of the binaries that are slowest. Again, these should be heavily cached so that you don't have to compile many times. Maybe it should grab the sources directly from the net, too... For instance, if you want to run ``grip'' it will automatically go to the net, get the grip source tree (and maybe a pre-compiled binary, too) and download it, and then start compiling/optimizing.
A woman drove her bicycle past several signs in a National Park that said ``No Bicycles!'' ``Hikers Only'' etc. When she drove off the trail and hurt herself she sued because the signs said that it was illegal to drive the bike, not dangerous.
I'm sick of all these slashdot weenies going around saying ``X sucks! We need to get rid of it!''
X is an amazing protocol.
No other windowing system that I have seen allows you to run graphical applications as flexibly as X. Run your application remotely and display it on your local machine (I always used the vacationing co-workers' Ultra 2's and Ultra 30's to do netscape and compilations rather than my U1). Run applications on your local machine, display them remotely. They have very good response. Even using a slowish connection.
Why is this? Because X is WELL DESIGNED. The model of the way X works is very cool. Not just for network applications, either. Many people say that X sucks because it slows down stuff done on the local machine. It is true that XFree86 can't display full-screen quake very well at high refresh rates, and part of that reason is due to going through the X protocol. It incurrs some amount of overhead. But for most applications, X is fine. And for those other applications, X can be extended in a legal and compatable way to bypass a lot of the network model for the local machine... which is what GLX is doing.
You also say that X doesn't have any support for AA fonts (or smoothed fonts). XFree86 doesn't. This is true. However, it is possible within the framework and architecture of X to have AA/smoothed fonts. And Dipsplay P(S|DF). And anything else that a display environment needs to do. And it can do it fast.
X IS NOT THE PROBLEM!!!
And, as we see processor speeds improve greatly over the next few years, both on the main board and on the graphics card, and we see busses not speed up at nearly the same rate, watch. You'll see that because of the modular, networked architecure of X, it will become lots more efficient when most of the X display code is running on your graphics card and your main processor doesn't need to send nearly so much information back and forth over the bus, because of the elegant architecture at its core.
If you want AA fonts, look into how to do it within X. Then do it. But don't whine about how Linux should adopt the Win or MacOS ways of doing things. Those ways suck. Fonts may look bad on my crappy monitor now under Linux, but it's stupid to drop an elegant architecture which can support nice fonts for one that sucks but has nice fonts.
Wrong! OS X is NOT a BSD clone. It is BSD-esque in some areas (mainly the microkernal, NOT the upper portions of the OS) and implements the BSD APIs, but it is fundamentally a different beastie
I think what the author meant was that you can't do BSD things (CLI stuff, configuring samba, writing bourne shell scripts) any easier than you can in BSD. So ``the power of having all your BSD stuff'' is not a net benefit over BSD.
Plus Aqua relies on proprietary technology that no open source initiative will ever license, and Adobe will never ever open source it. Never.
One if the significant reasons Apple went with ``display PDF'' instead of Display Postscript is that PDF doesn't have a lot of the wierd licensing considerations that PostScript does... so it would actually be a lot more reasonable to implement something compatable...
And, Display PostScript has been done and is somewhat understood... display PDF is just about the same thing.
Just because they work on the fruity OS for grandmothers and they aren't working for an open source company, doesn't mean they aren't incredibly talented, smart people.
No, but taking 20 years to come out with a computer with good dynamic memory allocation, virtual memory spaces, preemptive multitasking, and an architectural improvement in the display technology is proof enough that Apple didn't have talented, smart people until they bought NeXT. I mean, OS 9 is basically System 1 + Multifinder + TCP/IP + color, Really.
I liked Section 3 -- ``Provisions In Effect Until Full Implementation of the Plan of Divestiture''.
Choice parts of this included forbidding MS to coerce VARs, and I liked section iii -- ``Microsoft shall not restrict (by contract or otherwise, including but not limited to granting or withholding consideration) an OEM from modifying the boot sequence, startup folder, internet connection wizard, desktop, preferences, favorites, start page, first screen, or other aspect of a Windows Operating System Product (and goes on for specific things they can't do).''
This, in my opinion, may hurt MS more than the actual breakup. Forcing them to act fairly (and there's a compliance board they have to set up to assure this) may be the best thing that the Justice Dept could have done, even better than breaking them up.
Also, one of the nice things is that MS will have to disclose all their OS API's that other companies use -- no internal things that cause MS Apps to break! And my guess is that the development environments will be considered applications, so that means that all the libraries and such that MS uses will have to have an available API with the underlying OS -- meaning a better chance at obtaining compatability with things like WINE.
In short, I think the ruling is good -- least of all for the breakup. And my guess is that most of it won't be turned over on appeal.
While it's probably just the logo (like other people have mentioned) I wouldn't put it past MS to go up to the menu and hit ``smooth'' or whatnot. MS certainly has experience doctoring up photographs, and if they'd do it in a court of law I'm certain they'd do it for an advertisement.
Try Telocity. We (unfortunately) got BellSouth ADSL which has big problems. One of these problems is that they chose PPPoE or PPPoA(TM) as their protocols, so that Windows and Mac users could use their Dialup Adapters. Unfortunately, PPP is a connection based protocol, so the connection drops and the authentication server goes down lots (among other problems).
Other people in town (town == Atlanta, BTW) also use PPPoX, and also have big problems, though many are better than the awful BellSouth. None of these people support *nix, either.
A shining star in the whole area is Telocity. I have two friends now that have Telocity. A couple of days after BellSouth did the underlying phone testing and switching stuff (which took a while) Telocity had the DSL modem there and the service turned on. The DSL modem has a bunch of ports on it, so you can connect with your Ethernet port, USB port, parallel port, serial port, etc. It also comes with instructions on how to set your service up, on a nice little card that has three sections, from left to right: "Unix, Windows, Mac" I believe. You get a static IP address, and (from what I understand) if you move from any city to any other city that Telocity is in, you keep the same IP address. I thought that was cool. And there's no additional setup fee if you move. And the contract is month-to-month.
Anyway, it seems like they are extremely knowledgeable and have very high-quality service as a result. Though you still have stupid BellX that has to do a lot of the work, they will do the hassling for you, and once it's set up they have superior infrastructure and support.
The problem with trying to find out a solution to the game of chess is storing all the possible moves.
If you want to figure out a series of moves which could always lead to a victory (which may be possible) based on a single opening move, you could construct an algorithm that would play every possible game, and would figure out wich path did not allow the other player to win, no matter what he/she chose to do.
However, unlike distributed.net, where you know when you've found the answer and can throw away wrong answers, you have to save a whole lot of state for a brute-force chess game. The amount of storage necessary for all this state is probably prohibitive, even if we had millions of amazingly fast processors.
Slashdot Mirror
Of course, at $18 per CD, the RIAA stands to make lots of money off this.
And, using MP3s or streaming techniques violates the RIAA's right to be the sole source of acoustic art in the world.
It is all based on wording. If you agree with some of it, and not with others, cross it out, initial, and then sign the modified document. Or, add stuff in, initial, and sign.
Right, but even if you get an application to read your own catalog, it's still running as your userid. So it can only do as much damage as you could do normally.
Sure, you can change root-level programs to read from arbitrary files, but if you can change root-owned files then you can do anything you want in the system without having to make programs that do wierd things with buffer overflows and such.
Some would go farther to say that record labels are evil and the actual artists don't get any money anyway, blah blah blah, and with dissertations it's actually people stealing from the little guy, not the evil corporation. I don't know if that argument is valid or not, but just because napster (or gnutella, or whatever) allows people to distribute illegally copied materials doesn't make the technology itself illegal.
It's not illegal (or wrong) to control 100% of the market because you have a good (or popular) product. It is illegal (and wrong) to use your large market share to kill companies you don't like.
They found that it was bloated, poorly written, and poorly engineered. So they threw it all away and started over almost completely. Architecturally, the new Mozilla is much much better, and should be able to be more flexible over time than the old code tree. But it's taken a long time to get there. And it's not done yet. But it has many, many nice features.
Red Hat's Mission Statement / Corp Policy or whatever is that (from http://redhat.com/about) ``Red Hat shares all of its software innovations freely with the open source community under the GNU General Public License (GPL).'' Indeed, they tend to not talk about themselves in terms of a linux distribution company, but a company doing R&D, Integration, and distribution of Open Source software in general. Hence, the merger with cygnus.
The problem is with machines that block ICMP. Then you can't do MTU path discovery. This was the case with lots of ``secure'' sites, and when my ISP installed a transparent web proxy/cache they disabled ICMP on that too, meaning that every external web page stopped working.
Disk is worse. While RAM is, say, 10ns, a disk access is 10ms. That's 1 Million times slower. But we do the same thing... we can keep off the disk if we keep everything in RAM. And since we are multitasking and the disk is sooooo slow, if a process reads from disk we just take it off the processor and wait until the disk comes back with information before we will allow the process to come back on to the processor.
Now, since we're really smart, we do clever things to make sure that we have to go to Ram (or disk) as few times as possible. In cache, we load more than just the one byte or word that we want, we load the words next to it into cache as well, as that means that the next read from memory will (hopefully) be in cache already. This increases the penalty for going to RAM, but tends to pay off. Likewise, when we read a byte from disk, we load the 32K after it (or whatever we decide is appropriate) so that if we have to read the next byte it's in memory already.
Thus, we have really decreased the problems with the bottlenecks we do have. And it's very important... half the pins on modern processors are ground or power. That doesn't leave too many pins for I/O, and if you want to increase bandwith you usually want to increase the width of the bus in addition to bus speed. But unless you're Cray, it's really hard to run 1024 signals into your processors.
That's not to say that faster disks don't help, but you're not relying on your disk's speed every time you read or write a byte to your disk (thank goodness). Instead of trying to get the fastest disk, you may just want to increase the disk cache (IE, get more RAM). On my workstation, I can run Netscape and my terminals and text editor, and after I have opened them once or so, they just sit in RAM. When Netscape crashes and I have to start it up again, it pulls almost all of the program out of memory instead of having to go to disk. Instead of spending the money on a 10,000RPM Ultra Wide Fast Loose SCSI card and drive, I bought another 128M of RAM, and for most desktopish stuff (and a lot of heavier I/O, even) I can beat someone with less memory but a bigger disk.
The MTU is probably my biggest example of why I hate Windows and hate Mac more. On all the unix boxes, you type ``ifconfig eth0 mtu 1400'' or you can set the MSS on the default route (either fix the problem as they both do about the same thing). Under windows, you have to make a registry entry to fix the problem. With MacOS, you have to hunt your way around extremely little documentation and finally after reading mailing lists you find some random guy selling his little shareware tweaker.
That just seems like the MacOS way. Tons of annoying shareware for things that should just work and be in the distribution.
Which is different from what he was saying before. I've seen and heard Jobs spread lies about how superior the Mac was to anything. I've seen Apple tell everyone that they will be making their OS modern for years. I've seen them convince their customers that if they have to switch to Windows that they'll be dealing with IRQ conflicts and editing their config.sys all day long. How many times have you seen Mac people, even here on slashdot, use terms like ``plug-and-pray'' that they're convinced the Intel architecture is plagued with. I don't use windows much, but the last time I added a new video card to a Windows machine it worked just fine. And it's certainly been years since I dealt with an IRQ conflict. RedHat is amazing at figuring what hardware is added or removed from a system and configuring it.Not to mention most of the non-apple-done benchmarks I've seen put the G4s at a significantly lower performance level than even middle-of-the-road x86 chips. Try here, can't find some of the better comparisons I've seen.
Apple supports their own OS pretty well. Apple supports Windows in such a way as to make it annoying to use, as if they are penalizing them for choosing another OS. And while they say that many of the QT codecs are cross-platform, they really mean that they are available on Mac and Windows using the QuickTime Player from Apple Only. That's not a cross-platform standard in my opinion. Well, let's see. Why does Apple have these things? Is it because they're working towards compatability, or because they were having trouble getting hardware vendors to make specialized NuBus, LCBus, etc. cards when it only affected 10% of the potential market? Meanwhile, how much have they really done to make Macs interoperate well with PCs? Has apple put a SMB client in their OS yet? No! How about an NFS client? No! The only thing it ships with is Appletalk.Hmm... sort of like the NeXT monitor, that had power, video, sound, etc. all over the same cable.
Don't make it sound like it's innovative. Apple's just taking all the NeXT stuff and putting it in translucent smurf-puke blue containers. :-)
But Apple trying to converge with other OS's? That'll be the day. I think that Apple, as a corporation, is probably equally as evil as Microsoft, it's just that they haven't been in a position to do too much evil. Apple's not evil you say?
- Get other vendors to come up with clones, then refuse to license the OS after they finally get up and running.
- Support QuickTime under windows, but do it in such a way as to make the software as annoying as possible. Put your big logo up at the front every time it's launched. Make your own wierd control panel. Add yourself to the desktop. Claim to want to play everything, but not work on half the stuff. And the only reason they do it at all is because if they didn't support windows nobody would use QT at all.
- Don't support QT on anything but Windows
- Ship horrible mice and keyboards with your products, and then go with the line ``you can replace them with something else if you want!'' instead of shipping a reasonable product.
And, perhaps most importantly,- All the FUD and lies they spread. Steve Jobs claiming that you can't buy a faster computer than a mac with the G4 -- besides the fact that you could get an UltraSparc or Alpha, many PII's at the time beat it out in most things. Telling people that MacOS was technologically superior to everything. Telling them that it was compatable and flexible. You can't even change the MTU without third-party software.
Don't get me wrong. MacOS is very easy-to-learn. They put some interesting stuff in their boxes. But I wouldn't trust the company. I especially wouldn't trust the company to converge with anything. I'll be watching Apple try to stay as much in ultra-proprietary-land as they can.Motorola designed the HC11 specifically for automobile applications.
If you're sick of the assembly then get a different assembler... I seem to remember the HC16 assembly (which is much like HC11... I was writing rocket control softwarE) that I did was rather nice. Not quite as nice as PIC assembly, but...
NFS caching is mainly for reading. It's most helpful when I have large amounts of programs served by NFS, and want to cut down on network traffic.
And, I mean, we currently have NFS caching, but only for memory. You can't assign 2G of hard disk space for caching stuff read off of NFS.
The way read caching keeps consistency is using leases on the data... the server will contact the client if the data changes before the lease is up, and if the server can't reach the other client then the client is not assured correct data after the lease expires.
What we really need is Auto-Make in our operating system.
What I mean is, I should never have to generate a postscript file. Or a DVI file. If I want to print out or view the document I just edited, the OS should know that it needs to look for the TeX file and generate the DVI, then generate the postscript. It should cache them and delete them if I need the space.
Also, the operating system should generate binaries automatically from source. It should profile the binaries using spare CPU cycles and when the system is idle re-compile to try to optimize the parts of the binaries that are slowest. Again, these should be heavily cached so that you don't have to compile many times. Maybe it should grab the sources directly from the net, too... For instance, if you want to run ``grip'' it will automatically go to the net, get the grip source tree (and maybe a pre-compiled binary, too) and download it, and then start compiling/optimizing.
That's just my opinion, anyway.
Are we going to see NFS caching in Linux any time soon? It would make my life better...
A woman drove her bicycle past several signs in a National Park that said ``No Bicycles!'' ``Hikers Only'' etc. When she drove off the trail and hurt herself she sued because the signs said that it was illegal to drive the bike, not dangerous.
X is an amazing protocol.
No other windowing system that I have seen allows you to run graphical applications as flexibly as X. Run your application remotely and display it on your local machine (I always used the vacationing co-workers' Ultra 2's and Ultra 30's to do netscape and compilations rather than my U1). Run applications on your local machine, display them remotely. They have very good response. Even using a slowish connection.
Why is this? Because X is WELL DESIGNED. The model of the way X works is very cool. Not just for network applications, either. Many people say that X sucks because it slows down stuff done on the local machine. It is true that XFree86 can't display full-screen quake very well at high refresh rates, and part of that reason is due to going through the X protocol. It incurrs some amount of overhead. But for most applications, X is fine. And for those other applications, X can be extended in a legal and compatable way to bypass a lot of the network model for the local machine... which is what GLX is doing.
You also say that X doesn't have any support for AA fonts (or smoothed fonts). XFree86 doesn't. This is true. However, it is possible within the framework and architecture of X to have AA/smoothed fonts. And Dipsplay P(S|DF). And anything else that a display environment needs to do. And it can do it fast.
X IS NOT THE PROBLEM!!!
And, as we see processor speeds improve greatly over the next few years, both on the main board and on the graphics card, and we see busses not speed up at nearly the same rate, watch. You'll see that because of the modular, networked architecure of X, it will become lots more efficient when most of the X display code is running on your graphics card and your main processor doesn't need to send nearly so much information back and forth over the bus, because of the elegant architecture at its core.
If you want AA fonts, look into how to do it within X. Then do it. But don't whine about how Linux should adopt the Win or MacOS ways of doing things. Those ways suck. Fonts may look bad on my crappy monitor now under Linux, but it's stupid to drop an elegant architecture which can support nice fonts for one that sucks but has nice fonts.
I think what the author meant was that you can't do BSD things (CLI stuff, configuring samba, writing bourne shell scripts) any easier than you can in BSD. So ``the power of having all your BSD stuff'' is not a net benefit over BSD.
Plus Aqua relies on proprietary technology that no open source initiative will ever license, and Adobe will never ever open source it. Never.
One if the significant reasons Apple went with ``display PDF'' instead of Display Postscript is that PDF doesn't have a lot of the wierd licensing considerations that PostScript does... so it would actually be a lot more reasonable to implement something compatable...
And, Display PostScript has been done and is somewhat understood... display PDF is just about the same thing.
Just because they work on the fruity OS for grandmothers and they aren't working for an open source company, doesn't mean they aren't incredibly talented, smart people.
No, but taking 20 years to come out with a computer with good dynamic memory allocation, virtual memory spaces, preemptive multitasking, and an architectural improvement in the display technology is proof enough that Apple didn't have talented, smart people until they bought NeXT. I mean, OS 9 is basically System 1 + Multifinder + TCP/IP + color, Really.
And then we shall know that the end is near, for evil will have taken over the world.
Choice parts of this included forbidding MS to coerce VARs, and I liked section iii -- ``Microsoft shall not restrict (by contract or otherwise, including but not limited to granting or withholding consideration) an OEM from modifying the boot sequence, startup folder, internet connection wizard, desktop, preferences, favorites, start page, first screen, or other aspect of a Windows Operating System Product (and goes on for specific things they can't do).''
This, in my opinion, may hurt MS more than the actual breakup. Forcing them to act fairly (and there's a compliance board they have to set up to assure this) may be the best thing that the Justice Dept could have done, even better than breaking them up.
Also, one of the nice things is that MS will have to disclose all their OS API's that other companies use -- no internal things that cause MS Apps to break! And my guess is that the development environments will be considered applications, so that means that all the libraries and such that MS uses will have to have an available API with the underlying OS -- meaning a better chance at obtaining compatability with things like WINE.
In short, I think the ruling is good -- least of all for the breakup. And my guess is that most of it won't be turned over on appeal.
While it's probably just the logo (like other people have mentioned) I wouldn't put it past MS to go up to the menu and hit ``smooth'' or whatnot. MS certainly has experience doctoring up photographs, and if they'd do it in a court of law I'm certain they'd do it for an advertisement.
Other people in town (town == Atlanta, BTW) also use PPPoX, and also have big problems, though many are better than the awful BellSouth. None of these people support *nix, either.
A shining star in the whole area is Telocity. I have two friends now that have Telocity. A couple of days after BellSouth did the underlying phone testing and switching stuff (which took a while) Telocity had the DSL modem there and the service turned on. The DSL modem has a bunch of ports on it, so you can connect with your Ethernet port, USB port, parallel port, serial port, etc. It also comes with instructions on how to set your service up, on a nice little card that has three sections, from left to right: "Unix, Windows, Mac" I believe. You get a static IP address, and (from what I understand) if you move from any city to any other city that Telocity is in, you keep the same IP address. I thought that was cool. And there's no additional setup fee if you move. And the contract is month-to-month.
Anyway, it seems like they are extremely knowledgeable and have very high-quality service as a result. Though you still have stupid BellX that has to do a lot of the work, they will do the hassling for you, and once it's set up they have superior infrastructure and support.
If you want to figure out a series of moves which could always lead to a victory (which may be possible) based on a single opening move, you could construct an algorithm that would play every possible game, and would figure out wich path did not allow the other player to win, no matter what he/she chose to do.
However, unlike distributed.net, where you know when you've found the answer and can throw away wrong answers, you have to save a whole lot of state for a brute-force chess game. The amount of storage necessary for all this state is probably prohibitive, even if we had millions of amazingly fast processors.