CA's customers are almost exclusively large, mostly publicly held, companies and governments.
That was my point, actually. Big companies are potentially afraid of using linux because SCO might actually sue them. That would affect their bottom line, and they don't want to have to deal with defending themselves from a nonsense lawsuit just because CA deployed some linux servers/software for them. Individuals (non-corporate) who might otherwise use linux are really not worried about SCO suing them.
I don't know how many of you personally use metadata repository scanners or decision base implementation software, (at $30k per scanner, per license) but I'd bet they'll be okay without you.
Okay, now that's just creepy. I work for CA on their metadata repository scanner / ETL software. A DecisionBase product from (most recently) Platinum, no less. (this is, incidentally, a pretty small segment of CA's total business).
But yeah, you're exactly right, a bunch of slashdot readers aren't going to make a dent boycotting CA software, and the corporate customers aren't going to boycott.
I honestly think that for every company that pays SCO for that 'license' should be boycotted by the user community
This would not be productive. CA's minor contribution to SCO is not going to make the difference between SCO winning and losing their case. It might, however, make the difference between CA continuing to use, and sell, free of distraction, linux products to customers who might not feel comfortable using them otherwise. Which of these is better for the linux community?
Right now the iPod line-up is $300 - big 15 G, $400 - big 20 G, $500 - big 40 G, and now the new $250 - mini 4 G. To expand the line and continue to make money on it, they need new and exciting products; they can't just cut the price as their costs fall, because that will inevitably cut their profit margins down to nothing.
Ogg support aside, the first obvious answer is to keep doing what they've been doing for the past couple years, and introduce a $800 iPod with 80 gigs or something. Keep prices up, and just increase the sizes. This is even more pointless than the $250 4 gig iPod. 40 gigs is already so much music that no one really has any need for more than that, and no one would buy the new product. Result: Apple loses the mp3 player market to cheap generics.
Second, better answer: make it smaller, see how much more people will pay for a smaller unit. At $250, Apple can probably make a profit on each 4 gig unit. But more importantly, they can test the waters and see what the market for smaller-than-an-iPod players is. It's quite possible that the mini iPod will be a complete failure, but Apple needs to find that out. If it's simply priced too high for people, in 6-10 months they'll cut the price to $199 and see if there's a big increase in sales - in the mean time they'll sell a few at $250 to early adopters and Apple fanboys to recover development costs. But that's not the only variable. It could be that the original iPod was already small enough and no one cares about having anything smaller. It could be the the UI controls are unusably small for a lot of people. It could be people want a smaller product, but the mini iPod really isn't that much smaller. In which case they'll cube the product and tell their suppliers that they need an even smaller drive. Or maybe 4 gigs actually isn't enough space for enough people.
Assuming the mini iPod can be made a success (I have no doubt this will happen, but it may take price cuts before it does), in 2 years we'll see Apple with an iPod lineup something more like $200 - mini 8 G, $300 - mini 16 G, $400 - big 40 G. Result: Apple will have held on to prices, and therefore held on to their margin, and they'll have a profitable product line.
Seriously, that's enough IPs so that everybody on Earth can have their own unique set of IP addresses for each square nanometer on the surface of Earth. 128-bits is probably more than enough address space to handle everything that humans will ever want to address precisely. Unless we start giving out/32s to individual ISPs, we're not going to have a problem.
When IPv6 is found to be obsolete, it will not be because we've run out of addresses and are resorting to NAT to distribute them. It will probably be because we want more features out of IP unrelated to address space.
When Apple first introduced the $400, 5GB iPod, I had three reactions:
The firewire-only interface will seriously limit its market penetration
It costs too much for an mp3 player of that capacity
They should have used some sort of RAM (flash or otherwise) instead of a hard drive and they probably could have gotten better battery life and even better reliability, with the same size physical enclosure and the same price
It turns out Apple nicely dealt with the first two issues and they weren't that big a deal after all. I still wonder what you could do with a large-capacity portable non-disk based unit. At the time my thinking was that an iPod's size worth of SDRAM could easily handle the 5 gigs, and since the RAM can basically be as slow as you want, it would be cheap to produce. I have no idea how much battery it would require to power that much RAM, but I bet it would at least be comparable to periodically spinning up a hard drive while powering a smaller amount of RAM. The downside is that Apple would have to design a controller and interfaces that could deal with a big blob of RAM like that, while IDE controllers and interfaces are cheap.
It turns out I was wrong for completely different reasons. The hard drive is the better part of the iPod, technologically and financially. Before the iPod, a 5 GB drive of that size was a really, really new thing. The good news was someone else already developed it and Apple just had to buy them. And because tiny hard drives are useful for lots of applications, not just iPods, the market would make them bigger and cheaper for Apple to buy in the future. And so today Apple can offer a 40 GB iPod for only 20% more, and trivial additional R&D cost to Apple. There's a big market for RAM, too, but no one else cares about how much RAM you can fit in a smaller space, they care about making it faster mainly. One more thing: as hard drive capacities increase, Apple may be able to decrease the power usage by decreasing the rotational speed of the disk since you can get more data per rotation. RAM will likely use more power for more capacity.
Still, I wonder if this trend will reverse. It would be neat to see primary storage be done in RAM pools rather than hard disks, as people realize that their hard drives are more than big enough, but still the slowest parts of their system. I don't care what kind of musical taste you have, I have a hard time believing there's more than 80 gigs worth of 160kbps music out there which doesn't suck. I suspect that by the time that happens, though, people will be more interested in wirelessly streaming music to their iPods from storage located elsewhere, where it doesn't matter how big and power-consuming it is.
What do you care if I want my toaster to have a routable IP? The ONLY reason that's a bad idea is that IPv4 has created a scarcity of IPs. You just don't want to have your life inconvenienced by the transition to IPv6. Your argument is completely circular and pointless. And so we get stupid articles like this one by people like you who claim that we're not out of IP addresses and IPv4 isn't broken. The fact of the matter is that IPv4 is adequate to do a bunch of stuff and it's inadequate for a lot of other stuff. A lot of that other stuff is totally valid and your repeated presumption that it's all illegal or otherwise against my residential use contract isn't going to make it go away. Is it essential? Probably not - but how essential it is that everyone have cheap IPs is not in any way a function of the laziness of the network engineer who has to implement the protocol to support it. So let's not pretend that migrating to IPv6 is "pointless" because we won't "need" it for 20 years. Let's migrate as soon as we practically can and stop coming up with stupid excuses not to.
The reason there is a shortage is that people like you are cavalier about addresses.
I hardly see how. As I originally stated: I only get one IP from my ISP. I want more for non-commercial, non-bandwidth-intensive reasons which I consider legitimate.
IPv6 may be a good idea, I'm not debating that
Then why are you, as a network engineer, defending NAT as a solution, hmm? Because I'm a non-network-person, and I want people to have more IPs, I'm an obnoxious jerk who just wants to serve warez to kiddies from my home.
Uh, I don't know where you are from, but two phone numbers on one line has been possible and indeed sold in the United States for at least 45 years.
I am unaware of anywhere where a residential phone line consumer can request a second phone number for their residential line without buying a second line. Please provide a reference to availability and pricing of such a service.
Lemme guess- you're a web designer.
Software Engineer, sorry.
It boils down to this- IP's cost money. Pennies is ridculous and only shows you have no frame of reference to make these statements.
No shit IPs cost money - now. Please read the original comment you responded to. IPs cost money because we are already out of IPs. We are already out of IPs because we are still using IPv4. Hence the original article is a load of crap. I claim that if everyone used IPv6, ISPs would give out multiple IPs to regular users cheaply (except it wouldn't be pennies for each individual IPv6 address, it would probably be more like hundreds or thousands included in the basic cost of your service). This is because the fact of the matter is that IPs effectively cost NOTHING - as long as you're not stuck with IPv4.
Ok, explain to me in this wonderful world of free everything what your second machine with a routable IP will be doing.
Both computers will be simultaneously participating in the same (online game, online personal conferencing service, terminal sharing protocol, a million other things that don't matter).
Do you feel a second phone number from the phone company should be 2 cents a month too?
If you don't understand the difference between multiple phone numbers and multiple IPs you have completely missed the boat. For now, phone lines are burdened by a shitty but adequate protocol/routing system, and the internet is bound by a shitty and already inadequate IPv4 protocol. But I'll entertain your suggestion that some comparison could be made. Suppose phone numbers were allocatable independently from phone lines. Suppose that it were technically trivial to route two phone numbers over the same phone line. Suppose that there was no benefit (i.e. bandwidth of phone conversations) to having multiple phone numbers, other than that a person could call you on distinct numbers. And suppose that there was a virtually unlimited quantity of phone numbers available, as opposed to the current situation where (in the US) you get 10 digits, 3 of which are pretty sharply geographically bound, and 3 more of which are somewhat regionally bound - for example if there were, oh, I dunno, 2^128 phone numbers available in total. Then yes, I would say a reasonable cost for an additional phone number would be a few cents a month.
Macintosh may be a different platform, but still is a personal computer.
...or you could just use the language in the way it's defined, by a de facto standard, the way all natural languages are defined. And you could admit that it was "hackers" who defaced your website while you were eating some tasty "crackers" with cheese.
No one seemed confused at the presidential debate last night when the "Mac or PC" question was asked, insisting that a Mac was a PC too. So yeah, I'd say you're the one missing something.
"i guess when the users of a product don't hate the company that makes it, the honor system works."
I suspect with regard to Windows and product licensing serial numbers / activation, the converse is equally true. ("If the company uses the honor system, the users of its product don't feel like hating the company.")
I mean, if it's replacing the application, it's kinda going to have to quit it!
Not necessarily, but it's a reasonable expectation. Of course, it's a reasonable expectation that Apple didn't provide. When I installed the update I don't recall it giving me any indication that it was terminal.app that was going to be modified, let alone that it would be trying to quit terminal for me. They should have indicated both. Most Apple security updates require delayed reboots to finish installation, IIRC, so I was pleased to see that this one didn't at least.
First off, you are in no way a typical user if you have 6 comuters hooked up.
Perhaps I should have said "I am 3 or 4 typical users." One household, multiple people.
you need extra routable IP's to run a server. The IP isn't what you are being charged for, it's the bandwidth usage you will incur.
If by "server" you mean "something that listens and accepts connections on a well-known TCP or UDP port", then yes, I want to run servers. If by "server" you mean the more common definition, "something which serves content to other people", then no, not really. We must not give up and decide that running services (online games, ident, peer-to-peer protocols for conferencing and the like) is something to be reserved for "commercial" ISP customers. That is NOT the future.
Your one or two cents idea is foolish, and it porvesd that you don't understand this industry a bit.
I don't fault the ISP industry for charging so much; I fault the system for allowing it to be a reasonable way to do business. There is no reason more IPs = more bandwidth, and the system ought to accomodate a business model which makes it exceedingly practical to sell the two as independant variables, with "expensive" bandwidth and "cheap" IPs.
Routable IPs are for servers.
This is a really shitty attitude which serves only to defend IPv4's existence. There is no reason this condition should persist.
I am a more-or-less typical internet user. I have a cable modem from RCN for my household which happens to have 4-6 computers. Of course, right now I am using NAT. This is an incredibly lame solution for a number of reasons which have been discussed exhaustively here already.
RCN provides me with a connection, X bandwidth, and 1 IP.
My incremental cost of more IPs on the same connection and bandwidth is prohibitively high. (I would consider a penny or two per month per IP to be "reasonable" since each IP should have trivial overhead for the ISP)
Your concerns are well-founded. I discussed this long, long ago in this comment (yay google!). The only updates since that comment almost 3 years ago would be that 10.1/10.2 (1 and 2 years ago) broke compatibility with certain really old G3 PowerBooks, and that 10.3 (today!) breaks compatibility with "Old-World" Macs (more easily recognized as pre-USB Macs).
In summary, it's certainly uncomfortable that your OS upgrades and hardware upgrades come from the same source. However, Apple has generally shown itself to be a Good Guy in this regard, with the notable exception of MacOS X 10.0, which obviously made a number of sacrificies to be released to the public as early as it was.
I should also mention that XPostFacto has done a pretty good job of defeating Apple's unwillingness to support older hardware in the time since I wrote that comment.
I've got a number of pc's all around my house, and here at work - from 486's to PIV's. All running Linux or Windows XP (mostly Linux). All are very useable, given their expected purpose (ie 486==firewall).
I have for several years run a firewall on a 68040-based mac. This is a Mac from 1993, and it's still very "useable" in that regard. If I was running linux, I'd certainly expect that I'd be able to run the latest kernel on it. In reality, it runs MacOS 8.1 (from late 1997 or early 1998, I believe). And the reality is, if you're going to have low expectations of old hardware (I don't expect the Quadra to be usable as anything more than a router these days), there isn't much need to have modern, more usable software on it.
Security updates:
Just try to install the 10.2.8 security updates on 10.1 without purchasing 10.2. Then get back to me about who doesn't know what they're talking about.
Such a scheme results in higher new-prices, not lower upgrade-prices.
That is entirely an implementation detail to be left up to the implementer. There is no point in hypothesizing about what Apple would do because the fact is that Apple does not.
You assert that APPLE should give you a discount on an OS purchased for one machine when you buy another machine, when no company in any industry offers such a discount. Or are you under the impression that if you buy new tires for your car, then buy a new car, Ford (or whoever) should give you a discount equal to the price of the tires? Nice deal if you can get it - but just *try* to get it, from anybody.
Your car/tire analogy is thoroughly flawed. Software is a fundamentally different industry from that of automobiles. Yes, software consumers expect updates, and are not likely to do business with a company that never provides any. Even so, it fails because a) if I bought new tires (stereo, wheels, etc.) I would be free to install them on a new car for no additional cost; and b) my statement holds true for basically every piece of software ever save operating systems. Yes, if I buy some software, I am free to install it on my new machine, without paying the vendor for it - even if the computer vendor would have sold it to me. A good example is PC makers who offer to bundle Office with systems. If you own Office, you don't have to pay for it again. An ever better example is Windows. If I build my own PC, and I already own a copy of Windows, I'm under no obligation to pay Microsoft for another copy. This is only slightly harder buying a pre-built PC, but some will let you buy them OS-free. This is impossible with Apple because Apple has a monopoly on the supply of computers which run OS X (not that that makes Apple's monopoly comparable to Microsoft's).
We have come to expect upgrades of software. Apple has chosen to provide upgrades strictly on a two-tier basis: there are free upgrades (10.2.6, 10.2.8, etc.), and full-price upgrades (10.0, 10.2, 10.3). Full price is $129, which is a deal for full-price, but lousy for an upgrade. I think this model is too simple. Like many others, I think Apple should offer at least one of a discounted-upgrade program for people who paid for the OS recently, or a subscription model wherein I could pay a yearly fee and be guaranteed all updates.
There are more issues affecting Apple's choice not to price the update better.
For one, Apple has not done a very good job of making sure that security updates are available independent of OS version. Security updates must not be considered optional.
As a Mac user, I'd like to see Apple achieve the greatest market penetration possible with 10.3. This means making the barrier to entry as low as possible. I think 10.3 will be good for the Mac as a platform. If Apple chooses to maximize short-term profit rather than market penetration with 10.3, that's bad for market share, and that's bad for Mac software developers and potential developers.
There's also the issue of system upgrades. If I go out and buy a new Mac, which I'll hopefully do in the next several months, I am implicitly and effectively buying a new copy (license) of 10.3. Apple gives me no option to roll my separately-purchased license from my old computer forward to my new one (this is why I'd like to see an OS X subscription plan).
Finally, I'm still bitter that I paid for OS X 10.0 when it came out, and I installed it, but it was shelfware. On my not-very-old-at-the-time Mac, it was too slow to be usable, it was obnoxiously buggy, and it was missing features. OS X wasn't useable until 10.2, and even then it wasn't close to as pleasant to use as OS9 until I upgraded my CPU and video.
The philosophy of Mac OS X (and quite a few other operating systems, especially Unix-like ones) is that you should use as much RAM as you can.
This is a great philosophy until, of course, you've used as much RAM as you can and then you decide you want to load something new. It's perfectly reasonable to base your design on a user doing something like running a really small process over and over again - like, say,/bin/ls. But no one is ever going to quit Internet Explorer and start it again over and over, unless they're doing a benchmark to demonstrate how strong Apple's VM system is. Unfortunately, Apple chose to write for the synthetic benchmark rather than writing for a realistic usage case, I think.
A much more realistic case is that I'm using my memory loading web pages in Safari, I've got email going, itunes has been loading and playing mp3s for the past few weeks, etc., and then I decide I want to play a game. None of the game's data is cached, because it hasn't run recently. So the VM now has to scramble to find 200 MB of pages. Ideally the process of freeing unnecessary pages would be very fast, but it seems in practice (circa 10.2.x) it's going to have to swap a lot of stuff to disk first, which basically makes the whole process take twice as long as it could have if we had kept an ample supply of free memory available. OS X seems to have no problem finding "important" things to put in memory which can't be instantly freed, until there's very little free memory left.
Another complication of this is that it's pretty easy to fill up your root partition with swapfiles. Incidentally, this is terrible for disk performance. Like, making the system so slow that the cursor stalls for 10 seconds at a time. make -j comes to mind. And it's not particularly diligent about cleaning up swapfiles that are no longer used, which may not make any difference to the VM system, but it does to the filesystem, which still has to consider the disk space as in use.
One probably-related issue that comes to mind is that freeing memory seems to really take a long ass time even if you don't need it. The most glaring example of this is quitting Safari. Safari tends to "leak" memory (I don't know that it really does or if it actually has a legitimate use for tons of memory) over time/page loads, and occasionally I decide to quit it. Quitting Safari tends to stall my entire system for 10-30 seconds. [System: 500mHz G4 and over 500 MB of memory, YMMV] I really don't understand why this should be necessary. It seems to me that freeing all of a processes memory ought to be as simple as marking a lot of pages as dead, and it also seems to me that this entire process ought to be asynchronous with everything else since there's obviously nothing depending on those pages any more.
So what you're saying is, if I want a desktop with:
any PCI slots
upgradable video
more than 1 hard drive bay
more than two DIMM slots
a reasonable enclosure/case which I can stick under my desk (i.e. not a built-in monitor, because like the vast majority of buyers I already own one because I've owned a computer before)
I become a non-entry-level user, and suddenly I should pay $1300 instead of $1000, even after dropping the crappy 17" CRT? Okay buddy. Apple must know something about those non-entry-level users who don't want to pay over $1000 for a desktop, and their money, that Dell and Gateway don't.
You have no idea how to hire good programmers without getting a bunch of bad ones too.
You don't have any money to pay programmers until you've already developed something you can sell.
If you have enough time to do the job right, it's because you don't know what your customers want yet. By the time the job is done right, you'll need more time to make the job right again, because "right" has changed.
Seriously, the first one is the important one. You can't just get programmers who are properly trained and like to program. HR is notoriously incompetant. Even with good hiring practices, half of everyone you hire will be not nearly as good as your project requires. Pair programming is one way to address this: out of any given pair, one programmer will be more talented than the other. If you're lucky, this will improve the quality and quantity of output from the two programmers. If not, you may discover you need to go back to the hiring stange to replace the lesser of the two.
I've been playing this mud called AlteredReality for almost 6 years now. It's really good and has the most interesting codebase I've ever seen on a mud. It's based on ROM 2.4 but it doesn't play like a ROM at all. It has neat features like a completely persistent world, roundless fighting and a lot of real-time gameplay stuff. It has an original stat (10 primary stats)/class (they're up to 8 classes)/skill and group-skill system, and it gets rid of a lot of the ROM-lameness inherent in practices, trains, levels, etc. (although there are levels). Level-based PK (noloss) is allowed but it's not a serious PK mud. The NPCs fight mean but fair. Player run clans, and automated quests, arenas, mobprogs... all the stuff you'd expect from a modern MUD. =)
It also has some really cool features for the slashdot-geek, including ssh connections (ssh to port 4005, or telnet to port 4000), and a vt100 terminal mode which makes a mud client unnecessary. It's even got the latest openssh patch applied! The vt100 mode offers an inputline, prompt bar, line editing features, and there's built in speedwalks, etc.
It's got a really good social environment which is probably its other major selling point besides the nifty codebase. They have mudders from all over the world (Guam, Norway, etc. -- seriously). And if you're into building they have a full set of OLC functionality. Oh they have a website too, but it's not nearly as good as just logging into the game itself and making a character, perusing the help files, etc.
Maybe I'm just a san-serif hater (I like Times New Roman 14 (anti-aliased) for web pages), but how can anyone say that Arial is more readable than, uh, anything else? It scales horribly. At small sizes it's unreadable, and at large sizes it's just too plain.
Arial = Geneva with significantly worse spacing and uglier
Those are not good numbers. That's a net of less than 1%. I'm not saying they're dying, but those are not good numbers. Are you willing to buy a piece of a company with numbers like this?
For $1000 you can buy 0.000021% of the AAPL ($1000 / current AAPL Market Cap). 0.000021% of AAPL's current net tangible assets is about $835. In other words, you only need to have enough confidence in Apple to have a net growth of 20% over the period of your investment.
To make a totally invalid comparison, that same $1000 could buy you 0.00000037% of MSFT. Which only gets you about $195 of MSFT's net tangible assets. You have to have enough confidence in Microsoft for them to grow 413% over the period of that investment.
This completely ignores how profitable either company is per quarter or per year (P/E ratio, which is a totally valid benchmark), but any long-term estimate of that is much more speculative than the numbers I list above. Conclusion: AAPL is insanely cheap. If the stock market was rational, it would be priced much higher, and at this price you SHOULD be willing to buy a piece of a company with numbers like this.
This is significant - this is the first time anyone has promised a 745x G4 upgrade for Blue & White G3 PowerMacs - and thus the first time there's been a G4 for B&Ws faster than a 500 or 550 MHz 7400 (PowerLogix has a 550 MHz G4 ZIF but it comes with the cost of slower cache). This is a long time coming; but keep in mind that the product is only "expected in coming weeks" - so far it's still vapor.
At MacWorld New York last year Sonnet announced a few products, none of them for B&W G3s. I was there and I pestered the Sonnet booth people about this. They told me two things: B&W G3s are the largest untapped market for Mac CPU upgrades which they would of course love to have better offerings in, and that they couldn't comment on anything other than an announced product. I took this to mean that if it was feasible to put a 7450 in a B&W, they'd be selling one by now. A few months later I eBayed myself a Sonnet G4/500 ZIF, figuring I needed something to make my G3/400 B&W last until I could get a 970/G5 Mac from Apple (IMHO the current Apple G4 offerings are pretty lame with their underpowered bus and slow clock speeds, not to mention lack of modernized disk and expansion interfaces).
I've heard allusions to the possibility that the power requirements of a 7450 are too great for the PCI Motherboard and Power Supply used in the Yosemite/Yikes systems. [BTW, clock/bus ratio is not an issue, you can get a 10x ratio on a G4 and with a 100MHz bus there's plenty of room for more than 500MHz] I don't really understand this technically: my machine has enough juice to power 4-5 drives and 4 PCI cards, but it can't handle a faster processor? I'm hoping someone would like to comment on what the technical limitation was that made these upgrades take so long to come out, and how Sonnet plans to overcome them.
Slashdot Mirror
CA's customers are almost exclusively large, mostly publicly held, companies and governments.
That was my point, actually. Big companies are potentially afraid of using linux because SCO might actually sue them. That would affect their bottom line, and they don't want to have to deal with defending themselves from a nonsense lawsuit just because CA deployed some linux servers/software for them. Individuals (non-corporate) who might otherwise use linux are really not worried about SCO suing them.
I don't know how many of you personally use metadata repository scanners or decision base implementation software, (at $30k per scanner, per license) but I'd bet they'll be okay without you.
Okay, now that's just creepy. I work for CA on their metadata repository scanner / ETL software. A DecisionBase product from (most recently) Platinum, no less. (this is, incidentally, a pretty small segment of CA's total business).
But yeah, you're exactly right, a bunch of slashdot readers aren't going to make a dent boycotting CA software, and the corporate customers aren't going to boycott.I honestly think that for every company that pays SCO for that 'license' should be boycotted by the user community
This would not be productive. CA's minor contribution to SCO is not going to make the difference between SCO winning and losing their case. It might, however, make the difference between CA continuing to use, and sell, free of distraction, linux products to customers who might not feel comfortable using them otherwise. Which of these is better for the linux community?
This comment was specifically mentioned in a BusinessWeek article, and the comment wasn't even modded up?
Hmmm....Boggle.
Right now the iPod line-up is $300 - big 15 G, $400 - big 20 G, $500 - big 40 G, and now the new $250 - mini 4 G. To expand the line and continue to make money on it, they need new and exciting products; they can't just cut the price as their costs fall, because that will inevitably cut their profit margins down to nothing.
Ogg support aside, the first obvious answer is to keep doing what they've been doing for the past couple years, and introduce a $800 iPod with 80 gigs or something. Keep prices up, and just increase the sizes. This is even more pointless than the $250 4 gig iPod. 40 gigs is already so much music that no one really has any need for more than that, and no one would buy the new product. Result: Apple loses the mp3 player market to cheap generics.
Second, better answer: make it smaller, see how much more people will pay for a smaller unit. At $250, Apple can probably make a profit on each 4 gig unit. But more importantly, they can test the waters and see what the market for smaller-than-an-iPod players is. It's quite possible that the mini iPod will be a complete failure, but Apple needs to find that out. If it's simply priced too high for people, in 6-10 months they'll cut the price to $199 and see if there's a big increase in sales - in the mean time they'll sell a few at $250 to early adopters and Apple fanboys to recover development costs. But that's not the only variable. It could be that the original iPod was already small enough and no one cares about having anything smaller. It could be the the UI controls are unusably small for a lot of people. It could be people want a smaller product, but the mini iPod really isn't that much smaller. In which case they'll cube the product and tell their suppliers that they need an even smaller drive. Or maybe 4 gigs actually isn't enough space for enough people.
Assuming the mini iPod can be made a success (I have no doubt this will happen, but it may take price cuts before it does), in 2 years we'll see Apple with an iPod lineup something more like $200 - mini 8 G, $300 - mini 16 G, $400 - big 40 G. Result: Apple will have held on to prices, and therefore held on to their margin, and they'll have a profitable product line.
3*10^38 ought to be enough IPs for everybody. =)
Seriously, that's enough IPs so that everybody on Earth can have their own unique set of IP addresses for each square nanometer on the surface of Earth. 128-bits is probably more than enough address space to handle everything that humans will ever want to address precisely. Unless we start giving out /32s to individual ISPs, we're not going to have a problem.
When IPv6 is found to be obsolete, it will not be because we've run out of addresses and are resorting to NAT to distribute them. It will probably be because we want more features out of IP unrelated to address space.
When Apple first introduced the $400, 5GB iPod, I had three reactions:
- The firewire-only interface will seriously limit its market penetration
- It costs too much for an mp3 player of that capacity
- They should have used some sort of RAM (flash or otherwise) instead of a hard drive and they probably could have gotten better battery life and even better reliability, with the same size physical enclosure and the same price
It turns out Apple nicely dealt with the first two issues and they weren't that big a deal after all. I still wonder what you could do with a large-capacity portable non-disk based unit. At the time my thinking was that an iPod's size worth of SDRAM could easily handle the 5 gigs, and since the RAM can basically be as slow as you want, it would be cheap to produce. I have no idea how much battery it would require to power that much RAM, but I bet it would at least be comparable to periodically spinning up a hard drive while powering a smaller amount of RAM. The downside is that Apple would have to design a controller and interfaces that could deal with a big blob of RAM like that, while IDE controllers and interfaces are cheap.It turns out I was wrong for completely different reasons. The hard drive is the better part of the iPod, technologically and financially. Before the iPod, a 5 GB drive of that size was a really, really new thing. The good news was someone else already developed it and Apple just had to buy them. And because tiny hard drives are useful for lots of applications, not just iPods, the market would make them bigger and cheaper for Apple to buy in the future. And so today Apple can offer a 40 GB iPod for only 20% more, and trivial additional R&D cost to Apple. There's a big market for RAM, too, but no one else cares about how much RAM you can fit in a smaller space, they care about making it faster mainly. One more thing: as hard drive capacities increase, Apple may be able to decrease the power usage by decreasing the rotational speed of the disk since you can get more data per rotation. RAM will likely use more power for more capacity.
Still, I wonder if this trend will reverse. It would be neat to see primary storage be done in RAM pools rather than hard disks, as people realize that their hard drives are more than big enough, but still the slowest parts of their system. I don't care what kind of musical taste you have, I have a hard time believing there's more than 80 gigs worth of 160kbps music out there which doesn't suck. I suspect that by the time that happens, though, people will be more interested in wirelessly streaming music to their iPods from storage located elsewhere, where it doesn't matter how big and power-consuming it is.
I could have sworn I really did explain why I want multiple real IPs. But apparently you've already decided to ignore me. Obviously NAT "works", but it sucks. There's plenty of other comments in this discussion on the topic of NAT sucking if you don't want to believe me. Here's a nice list of things that NAT breaks which I took from the frizz's comment.
What do you care if I want my toaster to have a routable IP? The ONLY reason that's a bad idea is that IPv4 has created a scarcity of IPs. You just don't want to have your life inconvenienced by the transition to IPv6. Your argument is completely circular and pointless. And so we get stupid articles like this one by people like you who claim that we're not out of IP addresses and IPv4 isn't broken. The fact of the matter is that IPv4 is adequate to do a bunch of stuff and it's inadequate for a lot of other stuff. A lot of that other stuff is totally valid and your repeated presumption that it's all illegal or otherwise against my residential use contract isn't going to make it go away. Is it essential? Probably not - but how essential it is that everyone have cheap IPs is not in any way a function of the laziness of the network engineer who has to implement the protocol to support it. So let's not pretend that migrating to IPv6 is "pointless" because we won't "need" it for 20 years. Let's migrate as soon as we practically can and stop coming up with stupid excuses not to.
The reason there is a shortage is that people like you are cavalier about addresses.
I hardly see how. As I originally stated: I only get one IP from my ISP. I want more for non-commercial, non-bandwidth-intensive reasons which I consider legitimate.
IPv6 may be a good idea, I'm not debating that
Then why are you, as a network engineer, defending NAT as a solution, hmm? Because I'm a non-network-person, and I want people to have more IPs, I'm an obnoxious jerk who just wants to serve warez to kiddies from my home.
Uh, I don't know where you are from, but two phone numbers on one line has been possible and indeed sold in the United States for at least 45 years.
I am unaware of anywhere where a residential phone line consumer can request a second phone number for their residential line without buying a second line. Please provide a reference to availability and pricing of such a service.
Lemme guess- you're a web designer.
Software Engineer, sorry.
It boils down to this- IP's cost money. Pennies is ridculous and only shows you have no frame of reference to make these statements.
No shit IPs cost money - now. Please read the original comment you responded to. IPs cost money because we are already out of IPs. We are already out of IPs because we are still using IPv4. Hence the original article is a load of crap. I claim that if everyone used IPv6, ISPs would give out multiple IPs to regular users cheaply (except it wouldn't be pennies for each individual IPv6 address, it would probably be more like hundreds or thousands included in the basic cost of your service). This is because the fact of the matter is that IPs effectively cost NOTHING - as long as you're not stuck with IPv4.
Ok, explain to me in this wonderful world of free everything what your second machine with a routable IP will be doing.
Both computers will be simultaneously participating in the same (online game, online personal conferencing service, terminal sharing protocol, a million other things that don't matter).
Do you feel a second phone number from the phone company should be 2 cents a month too?
If you don't understand the difference between multiple phone numbers and multiple IPs you have completely missed the boat. For now, phone lines are burdened by a shitty but adequate protocol/routing system, and the internet is bound by a shitty and already inadequate IPv4 protocol. But I'll entertain your suggestion that some comparison could be made. Suppose phone numbers were allocatable independently from phone lines. Suppose that it were technically trivial to route two phone numbers over the same phone line. Suppose that there was no benefit (i.e. bandwidth of phone conversations) to having multiple phone numbers, other than that a person could call you on distinct numbers. And suppose that there was a virtually unlimited quantity of phone numbers available, as opposed to the current situation where (in the US) you get 10 digits, 3 of which are pretty sharply geographically bound, and 3 more of which are somewhat regionally bound - for example if there were, oh, I dunno, 2^128 phone numbers available in total. Then yes, I would say a reasonable cost for an additional phone number would be a few cents a month.
Macintosh may be a different platform, but still is a personal computer.
...or you could just use the language in the way it's defined, by a de facto standard, the way all natural languages are defined. And you could admit that it was "hackers" who defaced your website while you were eating some tasty "crackers" with cheese.
No one seemed confused at the presidential debate last night when the "Mac or PC" question was asked, insisting that a Mac was a PC too. So yeah, I'd say you're the one missing something.
"i guess when the users of a product don't hate the company that makes it, the honor system works."
I suspect with regard to Windows and product licensing serial numbers / activation, the converse is equally true. ("If the company uses the honor system, the users of its product don't feel like hating the company.")
I mean, if it's replacing the application, it's kinda going to have to quit it!
Not necessarily, but it's a reasonable expectation. Of course, it's a reasonable expectation that Apple didn't provide. When I installed the update I don't recall it giving me any indication that it was terminal.app that was going to be modified, let alone that it would be trying to quit terminal for me. They should have indicated both. Most Apple security updates require delayed reboots to finish installation, IIRC, so I was pleased to see that this one didn't at least.
First off, you are in no way a typical user if you have 6 comuters hooked up.
Perhaps I should have said "I am 3 or 4 typical users." One household, multiple people.
you need extra routable IP's to run a server. The IP isn't what you are being charged for, it's the bandwidth usage you will incur.
If by "server" you mean "something that listens and accepts connections on a well-known TCP or UDP port", then yes, I want to run servers. If by "server" you mean the more common definition, "something which serves content to other people", then no, not really. We must not give up and decide that running services (online games, ident, peer-to-peer protocols for conferencing and the like) is something to be reserved for "commercial" ISP customers. That is NOT the future.
Your one or two cents idea is foolish, and it porvesd that you don't understand this industry a bit.
I don't fault the ISP industry for charging so much; I fault the system for allowing it to be a reasonable way to do business. There is no reason more IPs = more bandwidth, and the system ought to accomodate a business model which makes it exceedingly practical to sell the two as independant variables, with "expensive" bandwidth and "cheap" IPs.
Routable IPs are for servers.
This is a really shitty attitude which serves only to defend IPv4's existence. There is no reason this condition should persist.
I am a more-or-less typical internet user. I have a cable modem from RCN for my household which happens to have 4-6 computers. Of course, right now I am using NAT. This is an incredibly lame solution for a number of reasons which have been discussed exhaustively here already.
RCN provides me with a connection, X bandwidth, and 1 IP.
My incremental cost of more IPs on the same connection and bandwidth is prohibitively high. (I would consider a penny or two per month per IP to be "reasonable" since each IP should have trivial overhead for the ISP)
Ergo, we are out of IPs already.
And that's what scares me about "going Mac".
Your concerns are well-founded. I discussed this long, long ago in this comment (yay google!). The only updates since that comment almost 3 years ago would be that 10.1/10.2 (1 and 2 years ago) broke compatibility with certain really old G3 PowerBooks, and that 10.3 (today!) breaks compatibility with "Old-World" Macs (more easily recognized as pre-USB Macs).
In summary, it's certainly uncomfortable that your OS upgrades and hardware upgrades come from the same source. However, Apple has generally shown itself to be a Good Guy in this regard, with the notable exception of MacOS X 10.0, which obviously made a number of sacrificies to be released to the public as early as it was.
I should also mention that XPostFacto has done a pretty good job of defeating Apple's unwillingness to support older hardware in the time since I wrote that comment.
I've got a number of pc's all around my house, and here at work - from 486's to PIV's. All running Linux or Windows XP (mostly Linux). All are very useable, given their expected purpose (ie 486==firewall).
I have for several years run a firewall on a 68040-based mac. This is a Mac from 1993, and it's still very "useable" in that regard. If I was running linux, I'd certainly expect that I'd be able to run the latest kernel on it. In reality, it runs MacOS 8.1 (from late 1997 or early 1998, I believe). And the reality is, if you're going to have low expectations of old hardware (I don't expect the Quadra to be usable as anything more than a router these days), there isn't much need to have modern, more usable software on it.
Just try to install the 10.2.8 security updates on 10.1 without purchasing 10.2. Then get back to me about who doesn't know what they're talking about.
Such a scheme results in higher new-prices, not lower upgrade-prices.
That is entirely an implementation detail to be left up to the implementer. There is no point in hypothesizing about what Apple would do because the fact is that Apple does not.
You assert that APPLE should give you a discount on an OS purchased for one machine when you buy another machine, when no company in any industry offers such a discount. Or are you under the impression that if you buy new tires for your car, then buy a new car, Ford (or whoever) should give you a discount equal to the price of the tires? Nice deal if you can get it - but just *try* to get it, from anybody.
Your car/tire analogy is thoroughly flawed. Software is a fundamentally different industry from that of automobiles. Yes, software consumers expect updates, and are not likely to do business with a company that never provides any. Even so, it fails because a) if I bought new tires (stereo, wheels, etc.) I would be free to install them on a new car for no additional cost; and b) my statement holds true for basically every piece of software ever save operating systems. Yes, if I buy some software, I am free to install it on my new machine, without paying the vendor for it - even if the computer vendor would have sold it to me. A good example is PC makers who offer to bundle Office with systems. If you own Office, you don't have to pay for it again. An ever better example is Windows. If I build my own PC, and I already own a copy of Windows, I'm under no obligation to pay Microsoft for another copy. This is only slightly harder buying a pre-built PC, but some will let you buy them OS-free. This is impossible with Apple because Apple has a monopoly on the supply of computers which run OS X (not that that makes Apple's monopoly comparable to Microsoft's).
We have come to expect upgrades of software. Apple has chosen to provide upgrades strictly on a two-tier basis: there are free upgrades (10.2.6, 10.2.8, etc.), and full-price upgrades (10.0, 10.2, 10.3). Full price is $129, which is a deal for full-price, but lousy for an upgrade. I think this model is too simple. Like many others, I think Apple should offer at least one of a discounted-upgrade program for people who paid for the OS recently, or a subscription model wherein I could pay a yearly fee and be guaranteed all updates.
There are more issues affecting Apple's choice not to price the update better.For one, Apple has not done a very good job of making sure that security updates are available independent of OS version. Security updates must not be considered optional.
As a Mac user, I'd like to see Apple achieve the greatest market penetration possible with 10.3. This means making the barrier to entry as low as possible. I think 10.3 will be good for the Mac as a platform. If Apple chooses to maximize short-term profit rather than market penetration with 10.3, that's bad for market share, and that's bad for Mac software developers and potential developers.
There's also the issue of system upgrades. If I go out and buy a new Mac, which I'll hopefully do in the next several months, I am implicitly and effectively buying a new copy (license) of 10.3. Apple gives me no option to roll my separately-purchased license from my old computer forward to my new one (this is why I'd like to see an OS X subscription plan).
Finally, I'm still bitter that I paid for OS X 10.0 when it came out, and I installed it, but it was shelfware. On my not-very-old-at-the-time Mac, it was too slow to be usable, it was obnoxiously buggy, and it was missing features. OS X wasn't useable until 10.2, and even then it wasn't close to as pleasant to use as OS9 until I upgraded my CPU and video.
The philosophy of Mac OS X (and quite a few other operating systems, especially Unix-like ones) is that you should use as much RAM as you can.
This is a great philosophy until, of course, you've used as much RAM as you can and then you decide you want to load something new. It's perfectly reasonable to base your design on a user doing something like running a really small process over and over again - like, say, /bin/ls. But no one is ever going to quit Internet Explorer and start it again over and over, unless they're doing a benchmark to demonstrate how strong Apple's VM system is. Unfortunately, Apple chose to write for the synthetic benchmark rather than writing for a realistic usage case, I think.
A much more realistic case is that I'm using my memory loading web pages in Safari, I've got email going, itunes has been loading and playing mp3s for the past few weeks, etc., and then I decide I want to play a game. None of the game's data is cached, because it hasn't run recently. So the VM now has to scramble to find 200 MB of pages. Ideally the process of freeing unnecessary pages would be very fast, but it seems in practice (circa 10.2.x) it's going to have to swap a lot of stuff to disk first, which basically makes the whole process take twice as long as it could have if we had kept an ample supply of free memory available. OS X seems to have no problem finding "important" things to put in memory which can't be instantly freed, until there's very little free memory left.
Another complication of this is that it's pretty easy to fill up your root partition with swapfiles. Incidentally, this is terrible for disk performance. Like, making the system so slow that the cursor stalls for 10 seconds at a time. make -j comes to mind. And it's not particularly diligent about cleaning up swapfiles that are no longer used, which may not make any difference to the VM system, but it does to the filesystem, which still has to consider the disk space as in use.
One probably-related issue that comes to mind is that freeing memory seems to really take a long ass time even if you don't need it. The most glaring example of this is quitting Safari. Safari tends to "leak" memory (I don't know that it really does or if it actually has a legitimate use for tons of memory) over time/page loads, and occasionally I decide to quit it. Quitting Safari tends to stall my entire system for 10-30 seconds. [System: 500mHz G4 and over 500 MB of memory, YMMV] I really don't understand why this should be necessary. It seems to me that freeing all of a processes memory ought to be as simple as marking a lot of pages as dead, and it also seems to me that this entire process ought to be asynchronous with everything else since there's obviously nothing depending on those pages any more.
- any PCI slots
- upgradable video
- more than 1 hard drive bay
- more than two DIMM slots
- a reasonable enclosure/case which I can stick under my desk (i.e. not a built-in monitor, because like the vast majority of buyers I already own one because I've owned a computer before)
I become a non-entry-level user, and suddenly I should pay $1300 instead of $1000, even after dropping the crappy 17" CRT? Okay buddy. Apple must know something about those non-entry-level users who don't want to pay over $1000 for a desktop, and their money, that Dell and Gateway don't.You have no idea how to hire good programmers without getting a bunch of bad ones too.
You don't have any money to pay programmers until you've already developed something you can sell.
If you have enough time to do the job right, it's because you don't know what your customers want yet. By the time the job is done right, you'll need more time to make the job right again, because "right" has changed.
Seriously, the first one is the important one. You can't just get programmers who are properly trained and like to program. HR is notoriously incompetant. Even with good hiring practices, half of everyone you hire will be not nearly as good as your project requires. Pair programming is one way to address this: out of any given pair, one programmer will be more talented than the other. If you're lucky, this will improve the quality and quantity of output from the two programmers. If not, you may discover you need to go back to the hiring stange to replace the lesser of the two.
I've been playing this mud called AlteredReality for almost 6 years now. It's really good and has the most interesting codebase I've ever seen on a mud. It's based on ROM 2.4 but it doesn't play like a ROM at all. It has neat features like a completely persistent world, roundless fighting and a lot of real-time gameplay stuff. It has an original stat (10 primary stats)/class (they're up to 8 classes)/skill and group-skill system, and it gets rid of a lot of the ROM-lameness inherent in practices, trains, levels, etc. (although there are levels). Level-based PK (noloss) is allowed but it's not a serious PK mud. The NPCs fight mean but fair. Player run clans, and automated quests, arenas, mobprogs... all the stuff you'd expect from a modern MUD. =)
It also has some really cool features for the slashdot-geek, including ssh connections (ssh to port 4005, or telnet to port 4000), and a vt100 terminal mode which makes a mud client unnecessary. It's even got the latest openssh patch applied! The vt100 mode offers an inputline, prompt bar, line editing features, and there's built in speedwalks, etc.
It's got a really good social environment which is probably its other major selling point besides the nifty codebase. They have mudders from all over the world (Guam, Norway, etc. -- seriously). And if you're into building they have a full set of OLC functionality. Oh they have a website too, but it's not nearly as good as just logging into the game itself and making a character, perusing the help files, etc.
While this is true - and fat binaries are a great thing, you are ignoring the fact that Apple desperately needs to replace Mach-O with something which is actually suited to performance on the PPC platform. My guess is that Apple will eventually replace it with something else (which will probably also support fatness).
Maybe I'm just a san-serif hater (I like Times New Roman 14 (anti-aliased) for web pages), but how can anyone say that Arial is more readable than, uh, anything else? It scales horribly. At small sizes it's unreadable, and at large sizes it's just too plain.
Arial = Geneva with significantly worse spacing and uglier
Those are not good numbers. That's a net of less than 1%. I'm not saying they're dying, but those are not good numbers. Are you willing to buy a piece of a company with numbers like this?
For $1000 you can buy 0.000021% of the AAPL ($1000 / current AAPL Market Cap). 0.000021% of AAPL's current net tangible assets is about $835. In other words, you only need to have enough confidence in Apple to have a net growth of 20% over the period of your investment.
To make a totally invalid comparison, that same $1000 could buy you 0.00000037% of MSFT. Which only gets you about $195 of MSFT's net tangible assets. You have to have enough confidence in Microsoft for them to grow 413% over the period of that investment.
This completely ignores how profitable either company is per quarter or per year (P/E ratio, which is a totally valid benchmark), but any long-term estimate of that is much more speculative than the numbers I list above. Conclusion: AAPL is insanely cheap. If the stock market was rational, it would be priced much higher, and at this price you SHOULD be willing to buy a piece of a company with numbers like this.
This is significant - this is the first time anyone has promised a 745x G4 upgrade for Blue & White G3 PowerMacs - and thus the first time there's been a G4 for B&Ws faster than a 500 or 550 MHz 7400 (PowerLogix has a 550 MHz G4 ZIF but it comes with the cost of slower cache). This is a long time coming; but keep in mind that the product is only "expected in coming weeks" - so far it's still vapor.
At MacWorld New York last year Sonnet announced a few products, none of them for B&W G3s. I was there and I pestered the Sonnet booth people about this. They told me two things: B&W G3s are the largest untapped market for Mac CPU upgrades which they would of course love to have better offerings in, and that they couldn't comment on anything other than an announced product. I took this to mean that if it was feasible to put a 7450 in a B&W, they'd be selling one by now. A few months later I eBayed myself a Sonnet G4/500 ZIF, figuring I needed something to make my G3/400 B&W last until I could get a 970/G5 Mac from Apple (IMHO the current Apple G4 offerings are pretty lame with their underpowered bus and slow clock speeds, not to mention lack of modernized disk and expansion interfaces).
I've heard allusions to the possibility that the power requirements of a 7450 are too great for the PCI Motherboard and Power Supply used in the Yosemite/Yikes systems. [BTW, clock/bus ratio is not an issue, you can get a 10x ratio on a G4 and with a 100MHz bus there's plenty of room for more than 500MHz] I don't really understand this technically: my machine has enough juice to power 4-5 drives and 4 PCI cards, but it can't handle a faster processor? I'm hoping someone would like to comment on what the technical limitation was that made these upgrades take so long to come out, and how Sonnet plans to overcome them.