Television content today is increasingly targeting dumb viewers. Advertisers are aware that intelligent viewers are not swayed by their advertising. To keep impressionable viewers watching, you need the kind of dumb content that draws them in. As a result, intelligent content is being pushed to the few premium providers that forgo traditional advertising.
It is something of a downward spiral. Content is stupid. Methods to access much of that content are still stupid. Savvy viewers quickly become frustrated with the fragmented paywalls, delayed releases and other obstacles, so they either pirate the content they want or simply go without. Why wait several months for the next season of Big Love to be released for streaming on Netflix when you can grab an HD MP4 of it from the Usenet or a Torrent site the day after it airs?
And the mini-sat and cable companies don't help things with their fucked up channel packages. To watch the handful of shows I still like, I'd have to subscribe to over $70/mo worth of channels. 98% of the content shown is little more than visual tripe. Why bother?
When the Boomers start dying off, traditional television as we know it will probably die with them. Maybe then we'll see a Renaissance in the television world. Until then, the people who came up with Retarded Guido TV, My Vagina is a Clown Car, Laugh at the Midgets Show and Lifestyles of Retarded Alaskan Politicians can all DIAF. So can the shitheads who watch it, too.
I agree that the forced trailers are awful. They were the main reason I stopped watching Blu-ray movies. And if you install a mod chip in your Blu-ray player that allows you to bypass the lock, you run the risk of ending up on a key blacklist.
Why would a DC-DC converter require rectification? You're using either pure DC or pulsed DC at that point. From your high voltage DC, wouldn't you just use a switching transistor to generate pulsed DC (which you smooth using caps) for the big-step down and then a linear regulator for smaller step-downs? Or is it just more efficient to convert back to AC and use a standard transformer coil?
I'm curious what voltage they run their DC circuits at.
When distributing power throughout your home or office, the AC vs DC argument isn't nearly as important as the voltage. The lower your voltage, the more current you need to transfer a watt of power. That means thicker gauge cabling to prevent overheating.
Obviously, the world has standardized on 110-120 and 220-240 for AC. However, medium distance DC is a mess of voltage standards, including 12VDC, 48VDC and others.
The best solution would be to run 240VDC to the wall and then have a step-down transformer in the form of a power brick or the like for your end devices. Since the brick no longer needs to be a rectifier, its construction is a lot simpler.
I disagree. Coal is cheaper because we don't force mining companies to clean their sites to a state of zero impact. We don't force generators to scrub their exhaust of 100% of all fly ash generated, to catalyze all nitrogen oxides or capture all carbon output. We don't force them to bury their fly ash, bottom ash and sludge a mile deep under a mountain.
If we did that, the cost of coal would go way up. Instead, what fly ash doesn't get expelled out the smoke stack is sold for use as filler in concrete, gypsum and fertilizers. Bottom ash also gets used as an aggregate. All this, even though such ash contains arsenic, barium, beryllium, cadmium, chromium, lead, mercury, molybdenum, radium, selenium, thorium, uranium and vanadium.
I think you meant to say fission reactors. AFAIK, fusion reactors that generate more power than they require to sustain a fusion reaction are still science fiction.
However, I do agree with your statement. Nuclear reactors [in combination with hydro and off-shore wind] would make for excellent base-load generation. The efficiency and safety of fission reactors has come a long way since the 1970s, which is the age of many reactors in the US.
The problem with nuclear is mostly image. People think of Chernobyl, which was an unsafe design even for its time. Thing is, you'd never have gotten a permit to build a Soviet RBMK turbine in the US, even in 1970 (the US required full containment and prohibited positive void coefficients). It is like suggesting that all cars today should be banned because somebody who didn't know how to drive got into a horrific crash in a '72 Yugo back in '86.
The other issue, waste disposal, is mostly an issue of NIMBYism. Vitrification does a fairly good job of making high-level waste stable for long-term storage. The bulk of our waste (90+%) is low-level crap like gloves, suits and the like - you probably get more radiation emissions from the radon escaping your granite counter-tops or seeping up through your basement. Even if the stuff leaks into the groundwater, we have strains of bacteria that we can pump into the water table that neutralizes it; it is being used at Hanford today.
Peak uranium isn't really an issue, either. The Japanese have found that you can extract uranium from seawater for about 2× the current cost of yellow cake on the commodities market. There is enough in the oceans to fuel us for 10,000 years. And remember that fuel is one of the smallest of costs of running a nuke plant. Then there is thorium, which is even more plentiful.
I recall a rumor that Sun Microsystems was once interested in purchasing Commodore. Sun was supposedly interested in the home and entry workstation marketplace, and wanted the Amiga line for themselves. I wonder if there was any truth to it.
Just imagine a SPARC based Amiga. That would have been interesting.
One problem that a lot of people have is that "one somebody" is the same for the world. You don't have a root.hints file broken down per country TLD. And even then, somebody has to maintain the root.hints file.
The other major problem is with the use of the.com,.edu,.net and.org TLDs. The United States never transitioned away from those domains in favor of its own.us TLD. As a result, the majority of organizations in the United States continue to use them. It wouldn't really be an issue except that organizations from around the world like to use them, too. So, are they de-facto extensions of the.us TLD, or are they extensions of a.world TLD?
So, if a root.hints file was created with a hint for each TLD, who would control the big four? An entity of (or delegated by) the US government, or an international entity such as the ITU? If it was the latter, how many US companies would rush to move their primary domains under a.us TLD?
FreeBSD has a centralized support system for everything from kernel drivers to userland binaries. Not so much with Linux.
FreeBSD has an excellent centralized configuration system. Almost all of my system configuration settings are stored in/etc/rc.conf. Under Linux distros, they tend to be scattered all over the place.
The FreeBSD ports and packages system is easier to use than APT or RPM.
Excellent Unicode support on a standard console (non-X) terminal using jfbterm.
Ability to run network services within separate sandboxes using the FreeBSD jail mechanism. Easier to setup than Linux-VServer.
I just downloaded Debian 6.0 the other day, and I was constantly frustrated at how much more difficult it was to setup a SOHO firewall with DNS, DHCP, HTTP and SMB services. I could see using a Linux kernel with a FreeBSD userland to gain some additional device drivers, but IMHO, the various userlands are a mess. I really wish that somebody would make the opposite of Debian/kFreeBSD.
It is. The x86 introduced micro-ops with the Pentium. The m68k introduced it with the 68060.
The whole idea is that more simplistic instructions allow for better pipelines. John Hennessy over at Stanford really pushed this while developing the MIPS architecture. RISC processors that use load/store instruction sets already meet this criteria. CISC processors retrofitted themselves by using complex instruction decoders that converted opcodes and their operands into micro-ops. At the cost of extra circuit complexity, you get the pipeline benefits of a RISC processor while keeping the more compact instructions of a CISC processor.
I've read in a few places that we may be one of the first around. Supposedly, heavy elements only came into abundant quantity around ten billion years ago. A much earlier universe couldn't have made our solar system.
OTOH, it would be an utter mindfuck to confirm that there is other life out there. Even moreso if it was intelligent. But it would be equally amazing if it turns out that we're the only ones because we came first.
Agreed. Take a real-world example: F5 Networks line of local area and wide area load-balancing products. Since the beginning, they have used some BSD variant as the base of their product (NetBSD 1.x, BSDi BSD/OS 3.1 and 4.1, FreeBSD 3.x) for BIG-IP and 3DNS. Their kernel patches and userland utilities were all closed source.
Eventually, they ran into a wall: BSDi was dying and FreeBSD didn't have the SMP performance they needed. Moving to GNU/Linux was an option, but the GPL would have forced them to move their code to userland (taking a heavy penalty while moving in and out of supervisor mode). Waiting for the SMP rewrite of FreeBSD that was underway was an option, but the timeline was murky and nobody knew how it would perform. So F5 Networks went with option three: create a custom closed-source microkernel called TMOS that sits above the Linux kernel and avoids the issue of GPL completely. My understanding was that it was a PITA to implement, but it leaves the Linux kernel untouched. As part of the rewrite, they renamed BIG-IP to LTM and 3DNS to GTM. That is their current offering today.
The Citrix Netscaler and Secure Computing's Sidewinder firewall are just a couple of products that continue to use FreeBSD because it was an easier path. AFAIK, F5 Network's path was a unique one.
Right, which is BS. If some guy can convince a 1000 college girls to strip naked every month and post their videos to his site for free, more power to him. The established pay sites are just going to have to come up with a compelling reason why their sites are better. If they can't, then they will fail.
My observation was just a general one and not specifically with regards to this lawsuit. But it is an observation with merit. In some circles, if you admit to paying for porn (as opposed to finding it for free), you're laughed out of the room. It is everywhere. But not all of it was supposed to be for free.
Porn is just one more market where people have a choice between something given away for free, something that costs money, and something that is supposed to cost money. The question revolves around how much item #3 hurts item #2, and how much item #1 will make the other two irrelevant.
The few times I've been to sites like in the article, over half of their material appeared to be scenes ripped from DVDs. I wonder if they have permission from the copyright holders for that stuff. Just seems to be yet another bunch of businesses that profit from stuff illegally uploaded and the amount of time it takes the right-holders to find it and yank it with a DMCA takedown notice.
Roughly half of the emails I receive for stuff I post in their for-sale section come from bots. I assume that they're trying to harvest email addresses for spam since Craigslist don't use a double-blind system. Almost all make no mention of the product in the body of the email and instead use a generic "is this item still available?" body.
These days, I add a statement to my postings that inform people emailing me to add the line "I'm not a bot!" or "I see purple people!" to the subject line of their email. Even if they're not a bot, I figure that if they can't follow simple instructions, they're not worth the time or effort to deal with.
I believe that an underspec'd firewall is most likely what they are referring to. Many people purchase firewalls based off of their raw bandwidth capacity. If you have an OC-12 ATM uplink to your ISP, basic logic used to suggest that you made sure that your firewall has at least an OC-12 or GigE port on the untrusted side.
But how many TCP SYN init packets can it parse per second? How many TCP connections can it handle before it runs out of memory? Does it treat embryonic connections different from a reaping standpoint than established connections? How many HTTP commands can it parse per second? All of a sudden, you have a lot more to worry about than bps throughput. You need to know the peak numbers of each in case you get slapped with a DoS attack.
Suddenly, that inexpensive 1Gbps firewall may not be enough. You might need to get a higher-end model, or you might need to bring in a Citrix or F5 load-balancer and spread the load.
That's actually my problem with the MOS 6510 - it was TOO similar to the MOS 6502. There were a bunch of bugs in the old 6502 core that had been around since it was released: weirdness with undefined op codes, wrapping index with direct page ops, unknown state of D flag after interrupt, unable to BRK after interrupt and so on.
The WDC 65C02 fixed all of these issues and then some. In fact, once you've gotten used to programming on a bug fixed 6502 variant (WDC 65C02, Rockwell 65C02S, Hudson 6580, WDC 65816 and CSG 65CE02), it is really painful to go back.
I would have really killed for one of the Rockwell 65C02 chips in the C64. In addition to the bug fixes, extra ops (DEA/INA, PHX/PLX, PHY/PLY, TSB/TRB), it also allowed you to relocate zero page and stack page to places other than page $00 and $01. Imagine being able to write your own interrupt routines using your own private stack and liberal use of zero page ops. GEOS would have run light years faster because you wouldn't need to save and restore the contents of the stack on a task switch.
You must not be aware of JIT style binary translators. I used to use x86 Windows binaries on my DEC Alpha running NT4 (and NT5 up till build 2000) under FX!32 all the time. Heck, I even ran Win32 games that used OpenGL.
Nothing prevents somebody from doing the same with ARM. In fact, I'm waiting for somebody to do that for WINE.
They say that they have working code for FreeBSD release-8. It makes me wonder if there is some relationship between Capsicum and FBSD's jails, or if FBSD is just being used because it is an environment of interest with the security/sandboxing community right now.
> Wrong. Original Amigas could present 4096 simultaneously in various ways.
You guys are cracking me up. I assume that you're referring to HAM and/or copper tricks. The limitations of HAM have already been described (great for some things, horrible for others). The problem with copper tricks is that you are still limited to a 32 color CLUT (or 32/32 with EHB6) for each scan line. It works very well for side-scrolling games where you have static areas of content staying within their specific vertical areas (such as grass or dirt at the bottom of your screen and sky or clouds at the top of your screen), but once you move to content that is a bit more dynamic, it behaves more and more like the base CLUT mode behind it.
>Amigas were able to produce 14 bit stereo sound and samplerates like 56khz.
Yet another hack. You're using the 6-bit volume register in conjunction with the 8-bit PCM data register; it isn't true 14-bit audio. The greatest precision is only with the quietest samples. As the volume of a sample gets louder, your precision starts to go back down to 8-bits. To make matters worse, the human ear is more sensitive to louder sounds, not quieter ones, so the most benefit with this trick is with medium loudness samples where they converge. And this hack locks out two of your PCM channels.
As for the higher sample rates, they were only with ECS and AGA chipsets using higher resolution screens with high-end processors running dedicated audio software.
All of these tricks also consume a fair amount of processor time. Copper tables are fine as long as you don't have to recalculate them with every refresh of the screen, but any sort of recalculation will impact your CPU. Simply watch any display program generate a S-HAM image from a true-color source as an example. Same goes with your 14-bit audio - unless the audio was already encoded to an ILBM file with both the 8-bit PCM and 6-bit audio, the processor has to generate it on the fly. And those 56KHz sample rates quickly become unattainable if you don't have enough processor time. Since the majority of Amigas had a 7MHz 68000...
While these tricks were neat at the time, they were still tricks. In the end, a 256 color global CLUT mode is easier to program for and work with than specialty modes that use delta offsets and scanline-specific CLUTs. A 12-bit or 16-bit PCM mode is easier than modulating a volume register and a PCM channel.
> And there's nothing like autoconfig anywhere yet.
I'm not even sure why you're bringing this up. Was Autoconfig simpler than PCI for setting device allocation? Yes. But then, PCI is a little more flexible.
No, I had an Amiga 4000 after my Amiga 3000. I missed my line doubler and onboard SCSI. I ended up getting a 4091 to keep my Quantum SCSI drives, but ran into the Buster bug. AGA was nice, but SVGA cards and SB16 cards were nicer. As the Amiga community started to die, I got a Cybervision64 in order to use Shapeshifter with native chunky screenmodes. In the end, I was using my A4000 as a Macintosh clone.
HAM is to graphics as DPCM is to audio. They are both technologies that use differential deltas to calculate a new value based off of an offset of the previous value. The problem with them is that you are limited by the resolution of your reference point and the range of your delta points. They also tend to dislike it when you inject a new reference point inside a chain of deltas. With HAM, that means that when an object moves across your beautiful HAM background, everything to the right of the object looks like a Greatful Dead concert barfed on your screen. That's why HAM was only used for niche applications.
As for the chosen memory model, AmigaOS couldn't use anything other than a simple flat model even if the original developers wanted to use a more elaborate system. The 68000 couldn't properly recover from memory bus faults, which are needed in order to detect access violations. This prevented the 68000 from being paired with an MMU - you needed a 68010 or better CPU. According to Haynie, 68020 processors were going for around $500 in bulk around 1985-86, so you're talking about using a 68010 if you wanted to keep your price close to sane.
You wouldn't have wanted to use a virtual paged memory model with the 68010 or 68020 anyway since there is a high cost in performing virtual-to-real memory translations. You'd kill your performance in a heartbeat. Worse, the 68451 was a pig of an MMU, and I'm not even sure if the 68551 was available yet (or at what cost).
You could have used a protected flat model with a custom MMU that only generated protection faults. Might have even been logic that you could have stuffed into Agnus. Since you're not doing translations, there's little if any overhead. But, as you mentioned, memory protection really wasn't something that designers had in mind for desktop OSes at the time. Which is why I think that if it wasn't introduced in the 1.x days, it really should have been introduced with 2.04 and the Amiga 3000.
Try listening to a Moog analog FM synthesizer or a high-quality digital FM synthesis program. There is a beauty to their sound that can never be replicated or equaled by simple PCM audio samples.
Granted, digital synthesis and digital sampling both have their strengths and weaknesses. But what bothers me is that Commodore had an inexpensive yet promising digital synthesizer in the MOS SID chip, and it went... nowhere. After Robert Yannes left the company the SID essentially died.
It is a shame, because had Commodore licensed frequency modulation synthesis from Yamaha, or had the licensing costs been too high, phase distortion modulation from Casio, they could have had a very powerful digital synth. They could have even outsourced it back to Yannes' new company.
Instead with the Amiga, they focused on PCM audio, because that’s what they got when they bought Amiga, Inc. and got Paula. Yet over the years, Paula remained the same, and so the Amiga’s PCM audio subsystem remained the same, even as other platforms moved on to 16-bit resolutions, higher sampling rates and adaptive compression.
Again, you see this lack of evolution. After OCS, Paula could have been broken into two – a basic I/O hub for serial, parallel, floppy and possibly even a single channel ATA controller in one chip, and a digital audio chip, possibly with basic sampling, FM synth and DSP effects in the other. Perhaps wishful thinking in hindsight, but hardly out of imaginational spirit of the original Amiga team.
Slashdot Mirror
Ahh yes, the first rule of alt.bin is not to talk about alt.bin.
It goes beyond this.
Television content today is increasingly targeting dumb viewers. Advertisers are aware that intelligent viewers are not swayed by their advertising. To keep impressionable viewers watching, you need the kind of dumb content that draws them in. As a result, intelligent content is being pushed to the few premium providers that forgo traditional advertising.
It is something of a downward spiral. Content is stupid. Methods to access much of that content are still stupid. Savvy viewers quickly become frustrated with the fragmented paywalls, delayed releases and other obstacles, so they either pirate the content they want or simply go without. Why wait several months for the next season of Big Love to be released for streaming on Netflix when you can grab an HD MP4 of it from the Usenet or a Torrent site the day after it airs?
And the mini-sat and cable companies don't help things with their fucked up channel packages. To watch the handful of shows I still like, I'd have to subscribe to over $70/mo worth of channels. 98% of the content shown is little more than visual tripe. Why bother?
When the Boomers start dying off, traditional television as we know it will probably die with them. Maybe then we'll see a Renaissance in the television world. Until then, the people who came up with Retarded Guido TV, My Vagina is a Clown Car, Laugh at the Midgets Show and Lifestyles of Retarded Alaskan Politicians can all DIAF. So can the shitheads who watch it, too.
I agree that the forced trailers are awful. They were the main reason I stopped watching Blu-ray movies. And if you install a mod chip in your Blu-ray player that allows you to bypass the lock, you run the risk of ending up on a key blacklist.
Why would a DC-DC converter require rectification? You're using either pure DC or pulsed DC at that point. From your high voltage DC, wouldn't you just use a switching transistor to generate pulsed DC (which you smooth using caps) for the big-step down and then a linear regulator for smaller step-downs? Or is it just more efficient to convert back to AC and use a standard transformer coil?
I'm curious what voltage they run their DC circuits at.
When distributing power throughout your home or office, the AC vs DC argument isn't nearly as important as the voltage. The lower your voltage, the more current you need to transfer a watt of power. That means thicker gauge cabling to prevent overheating.
Obviously, the world has standardized on 110-120 and 220-240 for AC. However, medium distance DC is a mess of voltage standards, including 12VDC, 48VDC and others.
The best solution would be to run 240VDC to the wall and then have a step-down transformer in the form of a power brick or the like for your end devices. Since the brick no longer needs to be a rectifier, its construction is a lot simpler.
I disagree. Coal is cheaper because we don't force mining companies to clean their sites to a state of zero impact. We don't force generators to scrub their exhaust of 100% of all fly ash generated, to catalyze all nitrogen oxides or capture all carbon output. We don't force them to bury their fly ash, bottom ash and sludge a mile deep under a mountain.
If we did that, the cost of coal would go way up. Instead, what fly ash doesn't get expelled out the smoke stack is sold for use as filler in concrete, gypsum and fertilizers. Bottom ash also gets used as an aggregate. All this, even though such ash contains arsenic, barium, beryllium, cadmium, chromium, lead, mercury, molybdenum, radium, selenium, thorium, uranium and vanadium.
I think you meant to say fission reactors. AFAIK, fusion reactors that generate more power than they require to sustain a fusion reaction are still science fiction.
However, I do agree with your statement. Nuclear reactors [in combination with hydro and off-shore wind] would make for excellent base-load generation. The efficiency and safety of fission reactors has come a long way since the 1970s, which is the age of many reactors in the US.
The problem with nuclear is mostly image. People think of Chernobyl, which was an unsafe design even for its time. Thing is, you'd never have gotten a permit to build a Soviet RBMK turbine in the US, even in 1970 (the US required full containment and prohibited positive void coefficients). It is like suggesting that all cars today should be banned because somebody who didn't know how to drive got into a horrific crash in a '72 Yugo back in '86.
The other issue, waste disposal, is mostly an issue of NIMBYism. Vitrification does a fairly good job of making high-level waste stable for long-term storage. The bulk of our waste (90+%) is low-level crap like gloves, suits and the like - you probably get more radiation emissions from the radon escaping your granite counter-tops or seeping up through your basement. Even if the stuff leaks into the groundwater, we have strains of bacteria that we can pump into the water table that neutralizes it; it is being used at Hanford today.
Peak uranium isn't really an issue, either. The Japanese have found that you can extract uranium from seawater for about 2× the current cost of yellow cake on the commodities market. There is enough in the oceans to fuel us for 10,000 years. And remember that fuel is one of the smallest of costs of running a nuke plant. Then there is thorium, which is even more plentiful.
I recall a rumor that Sun Microsystems was once interested in purchasing Commodore. Sun was supposedly interested in the home and entry workstation marketplace, and wanted the Amiga line for themselves. I wonder if there was any truth to it.
Just imagine a SPARC based Amiga. That would have been interesting.
One problem that a lot of people have is that "one somebody" is the same for the world. You don't have a root.hints file broken down per country TLD. And even then, somebody has to maintain the root.hints file.
.com, .edu, .net and .org TLDs. The United States never transitioned away from those domains in favor of its own .us TLD. As a result, the majority of organizations in the United States continue to use them. It wouldn't really be an issue except that organizations from around the world like to use them, too. So, are they de-facto extensions of the .us TLD, or are they extensions of a .world TLD?
.us TLD?
The other major problem is with the use of the
So, if a root.hints file was created with a hint for each TLD, who would control the big four? An entity of (or delegated by) the US government, or an international entity such as the ITU? If it was the latter, how many US companies would rush to move their primary domains under a
FreeBSD has a centralized support system for everything from kernel drivers to userland binaries. Not so much with Linux.
/etc/rc.conf. Under Linux distros, they tend to be scattered all over the place.
FreeBSD has an excellent centralized configuration system. Almost all of my system configuration settings are stored in
The FreeBSD ports and packages system is easier to use than APT or RPM.
Excellent Unicode support on a standard console (non-X) terminal using jfbterm.
Ability to run network services within separate sandboxes using the FreeBSD jail mechanism. Easier to setup than Linux-VServer.
I just downloaded Debian 6.0 the other day, and I was constantly frustrated at how much more difficult it was to setup a SOHO firewall with DNS, DHCP, HTTP and SMB services. I could see using a Linux kernel with a FreeBSD userland to gain some additional device drivers, but IMHO, the various userlands are a mess. I really wish that somebody would make the opposite of Debian/kFreeBSD.
It is. The x86 introduced micro-ops with the Pentium. The m68k introduced it with the 68060.
The whole idea is that more simplistic instructions allow for better pipelines. John Hennessy over at Stanford really pushed this while developing the MIPS architecture. RISC processors that use load/store instruction sets already meet this criteria. CISC processors retrofitted themselves by using complex instruction decoders that converted opcodes and their operands into micro-ops. At the cost of extra circuit complexity, you get the pipeline benefits of a RISC processor while keeping the more compact instructions of a CISC processor.
I've read in a few places that we may be one of the first around. Supposedly, heavy elements only came into abundant quantity around ten billion years ago. A much earlier universe couldn't have made our solar system. OTOH, it would be an utter mindfuck to confirm that there is other life out there. Even moreso if it was intelligent. But it would be equally amazing if it turns out that we're the only ones because we came first.
Agreed. Take a real-world example: F5 Networks line of local area and wide area load-balancing products. Since the beginning, they have used some BSD variant as the base of their product (NetBSD 1.x, BSDi BSD/OS 3.1 and 4.1, FreeBSD 3.x) for BIG-IP and 3DNS. Their kernel patches and userland utilities were all closed source.
Eventually, they ran into a wall: BSDi was dying and FreeBSD didn't have the SMP performance they needed. Moving to GNU/Linux was an option, but the GPL would have forced them to move their code to userland (taking a heavy penalty while moving in and out of supervisor mode). Waiting for the SMP rewrite of FreeBSD that was underway was an option, but the timeline was murky and nobody knew how it would perform. So F5 Networks went with option three: create a custom closed-source microkernel called TMOS that sits above the Linux kernel and avoids the issue of GPL completely. My understanding was that it was a PITA to implement, but it leaves the Linux kernel untouched. As part of the rewrite, they renamed BIG-IP to LTM and 3DNS to GTM. That is their current offering today.
The Citrix Netscaler and Secure Computing's Sidewinder firewall are just a couple of products that continue to use FreeBSD because it was an easier path. AFAIK, F5 Network's path was a unique one.
Right, which is BS. If some guy can convince a 1000 college girls to strip naked every month and post their videos to his site for free, more power to him. The established pay sites are just going to have to come up with a compelling reason why their sites are better. If they can't, then they will fail.
My observation was just a general one and not specifically with regards to this lawsuit. But it is an observation with merit. In some circles, if you admit to paying for porn (as opposed to finding it for free), you're laughed out of the room. It is everywhere. But not all of it was supposed to be for free.
Porn is just one more market where people have a choice between something given away for free, something that costs money, and something that is supposed to cost money. The question revolves around how much item #3 hurts item #2, and how much item #1 will make the other two irrelevant.
The few times I've been to sites like in the article, over half of their material appeared to be scenes ripped from DVDs. I wonder if they have permission from the copyright holders for that stuff. Just seems to be yet another bunch of businesses that profit from stuff illegally uploaded and the amount of time it takes the right-holders to find it and yank it with a DMCA takedown notice.
Roughly half of the emails I receive for stuff I post in their for-sale section come from bots. I assume that they're trying to harvest email addresses for spam since Craigslist don't use a double-blind system. Almost all make no mention of the product in the body of the email and instead use a generic "is this item still available?" body.
These days, I add a statement to my postings that inform people emailing me to add the line "I'm not a bot!" or "I see purple people!" to the subject line of their email. Even if they're not a bot, I figure that if they can't follow simple instructions, they're not worth the time or effort to deal with.
I believe that an underspec'd firewall is most likely what they are referring to. Many people purchase firewalls based off of their raw bandwidth capacity. If you have an OC-12 ATM uplink to your ISP, basic logic used to suggest that you made sure that your firewall has at least an OC-12 or GigE port on the untrusted side.
But how many TCP SYN init packets can it parse per second? How many TCP connections can it handle before it runs out of memory? Does it treat embryonic connections different from a reaping standpoint than established connections? How many HTTP commands can it parse per second? All of a sudden, you have a lot more to worry about than bps throughput. You need to know the peak numbers of each in case you get slapped with a DoS attack.
Suddenly, that inexpensive 1Gbps firewall may not be enough. You might need to get a higher-end model, or you might need to bring in a Citrix or F5 load-balancer and spread the load.
Three Mile Island, where the complaint was that there were too many alarms going off.
That's actually my problem with the MOS 6510 - it was TOO similar to the MOS 6502. There were a bunch of bugs in the old 6502 core that had been around since it was released: weirdness with undefined op codes, wrapping index with direct page ops, unknown state of D flag after interrupt, unable to BRK after interrupt and so on.
The WDC 65C02 fixed all of these issues and then some. In fact, once you've gotten used to programming on a bug fixed 6502 variant (WDC 65C02, Rockwell 65C02S, Hudson 6580, WDC 65816 and CSG 65CE02), it is really painful to go back.
I would have really killed for one of the Rockwell 65C02 chips in the C64. In addition to the bug fixes, extra ops (DEA/INA, PHX/PLX, PHY/PLY, TSB/TRB), it also allowed you to relocate zero page and stack page to places other than page $00 and $01. Imagine being able to write your own interrupt routines using your own private stack and liberal use of zero page ops. GEOS would have run light years faster because you wouldn't need to save and restore the contents of the stack on a task switch.
You must not be aware of JIT style binary translators. I used to use x86 Windows binaries on my DEC Alpha running NT4 (and NT5 up till build 2000) under FX!32 all the time. Heck, I even ran Win32 games that used OpenGL. Nothing prevents somebody from doing the same with ARM. In fact, I'm waiting for somebody to do that for WINE.
They say that they have working code for FreeBSD release-8. It makes me wonder if there is some relationship between Capsicum and FBSD's jails, or if FBSD is just being used because it is an environment of interest with the security/sandboxing community right now.
> Wrong. Original Amigas could present 4096 simultaneously in various ways.
You guys are cracking me up. I assume that you're referring to HAM and/or copper tricks. The limitations of HAM have already been described (great for some things, horrible for others). The problem with copper tricks is that you are still limited to a 32 color CLUT (or 32/32 with EHB6) for each scan line. It works very well for side-scrolling games where you have static areas of content staying within their specific vertical areas (such as grass or dirt at the bottom of your screen and sky or clouds at the top of your screen), but once you move to content that is a bit more dynamic, it behaves more and more like the base CLUT mode behind it.
>Amigas were able to produce 14 bit stereo sound and samplerates like 56khz.
Yet another hack. You're using the 6-bit volume register in conjunction with the 8-bit PCM data register; it isn't true 14-bit audio. The greatest precision is only with the quietest samples. As the volume of a sample gets louder, your precision starts to go back down to 8-bits. To make matters worse, the human ear is more sensitive to louder sounds, not quieter ones, so the most benefit with this trick is with medium loudness samples where they converge. And this hack locks out two of your PCM channels.
As for the higher sample rates, they were only with ECS and AGA chipsets using higher resolution screens with high-end processors running dedicated audio software.
All of these tricks also consume a fair amount of processor time. Copper tables are fine as long as you don't have to recalculate them with every refresh of the screen, but any sort of recalculation will impact your CPU. Simply watch any display program generate a S-HAM image from a true-color source as an example. Same goes with your 14-bit audio - unless the audio was already encoded to an ILBM file with both the 8-bit PCM and 6-bit audio, the processor has to generate it on the fly. And those 56KHz sample rates quickly become unattainable if you don't have enough processor time. Since the majority of Amigas had a 7MHz 68000...
While these tricks were neat at the time, they were still tricks. In the end, a 256 color global CLUT mode is easier to program for and work with than specialty modes that use delta offsets and scanline-specific CLUTs. A 12-bit or 16-bit PCM mode is easier than modulating a volume register and a PCM channel.
> And there's nothing like autoconfig anywhere yet.
I'm not even sure why you're bringing this up. Was Autoconfig simpler than PCI for setting device allocation? Yes. But then, PCI is a little more flexible.
> I was a serious user.
Yes, I can tell.
No, I had an Amiga 4000 after my Amiga 3000. I missed my line doubler and onboard SCSI. I ended up getting a 4091 to keep my Quantum SCSI drives, but ran into the Buster bug. AGA was nice, but SVGA cards and SB16 cards were nicer. As the Amiga community started to die, I got a Cybervision64 in order to use Shapeshifter with native chunky screenmodes. In the end, I was using my A4000 as a Macintosh clone.
HAM is to graphics as DPCM is to audio. They are both technologies that use differential deltas to calculate a new value based off of an offset of the previous value. The problem with them is that you are limited by the resolution of your reference point and the range of your delta points. They also tend to dislike it when you inject a new reference point inside a chain of deltas. With HAM, that means that when an object moves across your beautiful HAM background, everything to the right of the object looks like a Greatful Dead concert barfed on your screen. That's why HAM was only used for niche applications.
As for the chosen memory model, AmigaOS couldn't use anything other than a simple flat model even if the original developers wanted to use a more elaborate system. The 68000 couldn't properly recover from memory bus faults, which are needed in order to detect access violations. This prevented the 68000 from being paired with an MMU - you needed a 68010 or better CPU. According to Haynie, 68020 processors were going for around $500 in bulk around 1985-86, so you're talking about using a 68010 if you wanted to keep your price close to sane.
You wouldn't have wanted to use a virtual paged memory model with the 68010 or 68020 anyway since there is a high cost in performing virtual-to-real memory translations. You'd kill your performance in a heartbeat. Worse, the 68451 was a pig of an MMU, and I'm not even sure if the 68551 was available yet (or at what cost).
You could have used a protected flat model with a custom MMU that only generated protection faults. Might have even been logic that you could have stuffed into Agnus. Since you're not doing translations, there's little if any overhead. But, as you mentioned, memory protection really wasn't something that designers had in mind for desktop OSes at the time. Which is why I think that if it wasn't introduced in the 1.x days, it really should have been introduced with 2.04 and the Amiga 3000.
Try listening to a Moog analog FM synthesizer or a high-quality digital FM synthesis program. There is a beauty to their sound that can never be replicated or equaled by simple PCM audio samples.
Granted, digital synthesis and digital sampling both have their strengths and weaknesses. But what bothers me is that Commodore had an inexpensive yet promising digital synthesizer in the MOS SID chip, and it went... nowhere. After Robert Yannes left the company the SID essentially died.
It is a shame, because had Commodore licensed frequency modulation synthesis from Yamaha, or had the licensing costs been too high, phase distortion modulation from Casio, they could have had a very powerful digital synth. They could have even outsourced it back to Yannes' new company.
Instead with the Amiga, they focused on PCM audio, because that’s what they got when they bought Amiga, Inc. and got Paula. Yet over the years, Paula remained the same, and so the Amiga’s PCM audio subsystem remained the same, even as other platforms moved on to 16-bit resolutions, higher sampling rates and adaptive compression.
Again, you see this lack of evolution. After OCS, Paula could have been broken into two – a basic I/O hub for serial, parallel, floppy and possibly even a single channel ATA controller in one chip, and a digital audio chip, possibly with basic sampling, FM synth and DSP effects in the other. Perhaps wishful thinking in hindsight, but hardly out of imaginational spirit of the original Amiga team.