UFS support doesn't work that well because Mac OS X was designed to support both of its ancestors: OpenStep and Mac OS 9. Mac OS 9 applications rely on resource forks, file and creator types and case preservation and insensitivity, and they were often quickly ported to Carbon. No one wants to reconsider their app's fundamentals just to get it to run on a new OS; if they did, maybe we'd have a cleaner solution today.
Apple is moving towards ZFS, I just hope they'll start using it in Mac OS X client as well. All the neat features that *do* take up space (like revisions) and which people aren't used to can be easily turned off.
Most of Apple's reconsiderations of UNIX have been made to simplify or streamline what's there. Take launchd, which is their daemon that replaces rc.d and the startup system surrounding it. It was built to work with programs as they worked today. Upstart in Ubuntu was developed to be an entirely new design and work better and as a consequence probably does not work with completely unaltered programs. Tell me honestly: do you think people wouldn't have ragged on Apple for "being Apple" if they had done Upstart instead of launchd?
The problem isn't Apple making up new solutions to problems solved years ago, the problem is thinking these solutions can't be improved. Most (not all) of Apple's own problems in OS X with respect to being a UNIX citizen consists of compatibility junk that they're just now going to get around to dropping. (The newest version of Mac OS X manages to be certified as UNIX compliant, even if it's obviously not Linux certified since a different kernel is used.)
Finder was only "Cocoa-ized" to the point where Cover Flow is a Cocoa view. The rest is Carbon, it's just mostly a lot faster. The regression is actually, if you're to believe Apple design documents, a 'feature' since the view is now global. I guess it does solve the nearly impossible-to-predict mess that was there before to determine the view for a particular folder, but only in the same way that blowing up a car with a flat tire solves having a car with a flat tire. And for fucks sake, who decided we only needed three static columns in Spotlight? (All this metadata...)
I'm right with you on Core Input. Almost every Carbon-level API needs to be replaced and moved upstream. Objective-C if it provides helpful abstractions (or C++ if it is *really* performance-sensitive, like I/O Kit for drivers, which already exists), otherwise just clean, consistent and predictable CF-like C. I hope this is what Snow Leopard does that wasn't hot enough to reach the press release or sneak peek page.
I'm not sure, maybe because they said so very clearly in at least two quarterly earnings conference calls, and repeatedly in answer to other direct inquiries?
These two links are technically also from a weblog, albeit a different one, and you might deride that with the same argument as last time ("it's on a weblog" - which seems to be about as logical as placing every newspaper in the world, reputable or not, on equal footing simply because they're published in ways that are technically similar), but they also refer to two distinct Apple earnings conference calls, in which "Calendar 2008" and "10 million" are mentioned.
If you're still skeptical, you should easily be able to find what you regard as more reputable sources transcribing the same calls to confirm or deny.
Automator is nothing like Hypercard except for the part where it does something somehow related to scripting. (And I'll be damned if I'll compare anything related to scripting to Hypercard.)
You could build databases and small Flash-like games with Hypercard (inevitably someone will point out how an early version of Myst, maybe the demo, was written in Hypercard, so it might as well be me). You can build workflows - scripting pipelines - with Automator, and in specific cases you can do it much easier and better than in Hypercard since it's aimed precisely for that.
So no, Automator really is nothing like Hypercard. Hypercard's very little like anything except perhaps for Flash and Visual Basic. (If this was what you were saying or trying to get at, it wasn't readily apparent.)
LINQ's order would be logical even if IDEs weren't even involved, since it leads you to define every variable before you use them, which is the default in every other place you use variables. If they had used SQL's order, this basic logical assumption would have been broken. (I'm not necessarily disagreeing with anything the parent said, I just felt this had to be pointed out.)
They didn't drop it out of spite as such. A lot of the existing VBA code was hand-rolled PowerPC ASM *and* based on a since-surpassed VBA version, in combination with an understaffed department for this sort of support. "Running to stand still" wouldn't really cover it, although I'm starting to wonder what has changed since; maybe it's just become more tenable to port the Windows version, or maybe people have been hired.
It's interesting to compare this to Microsoft's own research and actual usage data backing the entire redesign effort, which showed several real problems and which led the work.
I'm not all that in love with most other parts of Office 2007, especially not the underlying politics and company guiding the OOXML bullshit and anything-open-has-cooties thinking. But I have read up on how the redesign happened, point by point, and I can't fault them for not doing their homework.
It's a solid piece of engineering and craftsmanship (if you remove the horrendous branding like the "Office button"), but it's hard to judge the merits of the interface based on the first iteration of it, plagued by lack of customization and immense culture shock in anyone who sees it. The application of the interface to the programs might not have worked so well in practice as they thought it would in theory, but I think it's also fairly clear that "stay the course" would not have worked that well for that long.
There are plenty of beefs with the "crapware barrier" theory, although it sounds like reasoning Apple would apply. For one, the bucks prevent people who don't otherwise want to put out apps from making small, quick apps entirely for themselves, for personal use. Maybe Apple's just going to take their lumps or (inadvertently or not) nudge-nudge, wink-wink developers towards jailbreaking/pwning for those cases.
Furthermore, don't discount open source software (and I mean the open source construction process, not the Free Software licensing model and motive). To take one example, Adium on Mac OS X is one of the best citizens on that platform, away and beyond most commercial apps. You and Apple are right that this takes strong guidance, and per Sturgeon's law the vast majority aren't up to this. But the $99 fee that "they" can swing in the case of Firefox isn't per product or company, it's per *person*.
You're exactly right that it's not the absence of some crappy or unintuitive software that I'm going to miss on the iPhone; it's the presence of crappy or unintuitive software produced by people with money, and absence of potentially great software that could never be produced.
I'm expecting the community to hack this by creating distribution houses - something like software "labels" - that you can go through. But it doesn't solve simple stuff like beta testing without sending everyone your source code and having them download your SDK. I think Apple's greed and/or want for control (recall how old versions of APE triggered non-loading 10.5 installs six months back) has them overlook real problems.
The masses don't care one whit about the technology or licensing behind their apps, but the big deal with iPhone development is that to be able to test anything on a real phone "legitimately", you have to pay $99 for a developer license that's hitherto only available for US developers. For a "friggin $500 phone" (even if the 8 GB model is nowadays $399, so $400), this is *not* a big deal if you're planning on selling software, but a big deal if you're planning on releasing free software.
On no other mobile platform that I know of do you have to pay just to be able to try it out on a real device or just to be able to release freeware, and with the iPhone we're talking, in terms of user interface, the most technically advanced mobile platform. No wonder people are a bit upset that they can't use it. There's no accelerometer and no location awareness in the Simulator.
The "US only developers" thing is obviously to limit the test group to a manageable size until the SDK is really ready. We don't yet know what of the other restrictions are here to stay (I'm betting code signing and the App Store-only distribution are the most ingrained), but I really hope that it's a ruse and that most of the real brain farts will have been removed by WWDC, when the final SDK drops.
Yes, accounting reasons is probably why they did it, although not *really* because then they would have charged just a few bucks like with the N enabler.
If the update is to be seen as a dry run, I still think it's curious since most of this process is not at all how applications will be delivered with the SDK. (I know as little as anyone, but I have a very hard time imagining new firmwares for delivering third party software.) And my takeaway from this is that the update was not a dry run, or a dry run of a specific component still applicable.
I could perhaps imagine that people would get a Mac mini to develop iPhone applications, but it sounds absurd that Apple would have people who already own Macs buy new multitouch-equipped laptops. (And you can't even see the screen, obviously, on the trackpad.)
Signed applications means that the applications have a cryptographic signature attached to them with *any* trust root, not that they have to be attached to any specific trust root. There's a leap from "applications have to be signed" to "applications have to be signed by Apple" or the more likely "applications have to be signed by an authority whose certificate is trusted by Apple". But let's say that happens: unless all development happens inside a simulator in software (and good luck testing multi-touch then), I definitely think that there'll be a way to run "untrusted" apps, and this way will be exploited to run free apps. I think Apple knows this.
The iPod touch update was curious - the apps were already in the new firmware, and the update just "unlocked" them. (The update weighs in at 9 KB.) Since people won't get to download new firmware every time they get an app, this doesn't confirm much, although I agree that it was probably a dry run of some component in the whole scheme, most likely signing.
It's interesting that everyone takes for granted that "getting from iTunes and then syncing it over" will be the only way to get apps. It's likely to be one of the ways, but Apple has revealed nothing. It's all speculation so far.
I doubt Apple is going to host any freeware programs that people write out of the goodness of their hearts.
They already host downloadable Dashboard widgets and provide links to all sorts of software on their site and host the world's biggest podcast directory at no fee for anyone, producers or users. I don't see how helping to host applications that could solve every non-hardware related aspect ("3G!") of their product would be *bad* for them, even if some of those applications were free.
I expect to see some way that Apple will help people sell their apps if they do end up with some sort of iTunes app store, but one approach doesn't rule out the other, especially since it likely won't be that easy to get access to their payment/transaction system.
The latest official word on this that I can recall says: "The main idea behind the Optimus Maximus key design is that the part with the display is fixed, while the transparent cap is moving, pressing a Cherry switch underneath[..]"
* Visual voicemail. New functionality on this order demands special implementation.
I had a similar feature via the web from CollegeClub back in 2000. Voicemail looked like e-mail with audio attachments. You could also check your e-mail and voicemail by calling their 888 number.
Most VOIP providers I've dealt with also have visual voicemail via an app or a website. The phone has a browser and an e-mail client. Visual voicemail shouldn't require carrier exclusivity.
Shouldn't, no. Imagine Apple courting carriers to implement stuff on their network so that they could deliver visual voicemail to a network that the phone isn't bound to. I'm not saying I wouldn't like ubiquitous visual voicemail, but it wasn't implemented before in that way and I don't think any carrier would take them up on the offer.
* Unlimited data. Regardless of 3G or EDGE, data on *any* cell phone that's not specifically a 3G modem tends towards ridiculous fees. If Apple had released an unlocked iPhone and asked for unlimited data plans, the carriers would laugh and ask if they also wanted a pony.
I have unlimited data for $15 from Sprint on a platform where you can actually do useful things with that data (Treo 650).
I can run ssh and a powerful e-mail client. I can take the full-sized SD card out of my camera and stick it in my phone. Then I can e-mail the pictures. These are things that the newer, more expensive, carrier-exclusive iPhone can't do.
Fair point - unlimited data has been something I haven't seen offered myself unless you spring for a business plan or go for the full flat rate (unlimited everything) plan.
* To gain a foothold in the total clusterfuck that is the US mobile market.
This is the only valid point in this list. AT&T exclusivity got the iPhone out there. Now that it's an established product, Apple should start selling it unlocked at a price that covers the kickback AT&T pays them on the contract.
Couldn't agree more, I just hope their "multi-year exclusive" isn't that "multi".
Of course they can't. It is dumb. Your argument is correct.
In this specific instance, I happen to agree that there should be unlimited data plans, and there aren't that many that aren't also focused on "unlimited anything" (flat rate), at ridiculous prices.
There are three reasons I see for Apple ever wanting carrier exclusivity in the first place:
* Visual voicemail. New functionality on this order demands special implementation.
* Unlimited data. Regardless of 3G or EDGE, data on *any* cell phone that's not specifically a 3G modem tends towards ridiculous fees. If Apple had released an unlocked iPhone and asked for unlimited data plans, the carriers would laugh and ask if they also wanted a pony.
* To gain a foothold in the total clusterfuck that is the US mobile market.
They're all pretty sad reasons, which doesn't mean they aren't probable. And it just sucks that Apple took their strategy to countries where it's not really needed except to remain consistent with the AT&T agreement.
That makes some sense. They've lost the road warrior market if they don't have swappable batteries, but with LED backlighting, most things on the motherboard, a 1.8" drive (and SSD if you choose so, I suppose) and a minimum of extra ports (and optical drives) to power, they may be able to crank up the battery life to an acceptable level. Rumor has it that the advertised 5 hours were recorded with Wi-Fi on, as opposed to the standard "every feature off, RAM disk, screen minimum brightness, CPU napping and downclocked and fair wind; oh, and it never actually *did* anything during that time, this is just a theoretical maximum" industry fare. Then again, road warriors would probably like to see bigger hard drives.
Because when they don't need to make a hatch for the battery, they could make it thinner.
Seriously: the reason MacBook Air exists is because it's light, and because it's ridiculously thin. It's also interesting because of the trackpad and its multi-touch gestures, and because it has slightly different tradeoffs when compared to other subnotebooks which makes it a fairly unique product - MacBook Air has a faster CPU and normal-sized screen and keyboard, virtually every subnotebook has more features and ports in every other aspect, but piddly screens and keyboards.
The reason MacBook Air exists is because it's a unique proposition in the market (per the above), not because it's really convenient to service yourself or fast or extensible. There are other products for that.
Slashdot Mirror
UFS support doesn't work that well because Mac OS X was designed to support both of its ancestors: OpenStep and Mac OS 9. Mac OS 9 applications rely on resource forks, file and creator types and case preservation and insensitivity, and they were often quickly ported to Carbon. No one wants to reconsider their app's fundamentals just to get it to run on a new OS; if they did, maybe we'd have a cleaner solution today.
Apple is moving towards ZFS, I just hope they'll start using it in Mac OS X client as well. All the neat features that *do* take up space (like revisions) and which people aren't used to can be easily turned off.
Most of Apple's reconsiderations of UNIX have been made to simplify or streamline what's there. Take launchd, which is their daemon that replaces rc.d and the startup system surrounding it. It was built to work with programs as they worked today. Upstart in Ubuntu was developed to be an entirely new design and work better and as a consequence probably does not work with completely unaltered programs. Tell me honestly: do you think people wouldn't have ragged on Apple for "being Apple" if they had done Upstart instead of launchd?
The problem isn't Apple making up new solutions to problems solved years ago, the problem is thinking these solutions can't be improved. Most (not all) of Apple's own problems in OS X with respect to being a UNIX citizen consists of compatibility junk that they're just now going to get around to dropping. (The newest version of Mac OS X manages to be certified as UNIX compliant, even if it's obviously not Linux certified since a different kernel is used.)
Finder was only "Cocoa-ized" to the point where Cover Flow is a Cocoa view. The rest is Carbon, it's just mostly a lot faster. The regression is actually, if you're to believe Apple design documents, a 'feature' since the view is now global. I guess it does solve the nearly impossible-to-predict mess that was there before to determine the view for a particular folder, but only in the same way that blowing up a car with a flat tire solves having a car with a flat tire. And for fucks sake, who decided we only needed three static columns in Spotlight? (All this metadata...)
I'm right with you on Core Input. Almost every Carbon-level API needs to be replaced and moved upstream. Objective-C if it provides helpful abstractions (or C++ if it is *really* performance-sensitive, like I/O Kit for drivers, which already exists), otherwise just clean, consistent and predictable CF-like C. I hope this is what Snow Leopard does that wasn't hot enough to reach the press release or sneak peek page.
Exactly.
Why would Apple set the bar so high?
I'm not sure, maybe because they said so very clearly in at least two quarterly earnings conference calls, and repeatedly in answer to other direct inquiries?
These two links are technically also from a weblog, albeit a different one, and you might deride that with the same argument as last time ("it's on a weblog" - which seems to be about as logical as placing every newspaper in the world, reputable or not, on equal footing simply because they're published in ways that are technically similar), but they also refer to two distinct Apple earnings conference calls, in which "Calendar 2008" and "10 million" are mentioned.
If you're still skeptical, you should easily be able to find what you regard as more reputable sources transcribing the same calls to confirm or deny.
http://www.macrumors.com/2007/01/17/apple-posts-1-billion-in-profit-1q-2007-and-financial-call-notes/
http://www.macrumors.com/2007/10/22/apple-4q-2007-results-conference-call-6-22-billion-revenue-904-million-prof/
Well, no. Apple's confirmed that they intended "10 million in 2008". See: http://www.alleyinsider.com/2008/3/mystery_solved__what_apple_s_10_million_iphones_goal_means__aapl_
Automator is nothing like Hypercard except for the part where it does something somehow related to scripting. (And I'll be damned if I'll compare anything related to scripting to Hypercard.)
You could build databases and small Flash-like games with Hypercard (inevitably someone will point out how an early version of Myst, maybe the demo, was written in Hypercard, so it might as well be me). You can build workflows - scripting pipelines - with Automator, and in specific cases you can do it much easier and better than in Hypercard since it's aimed precisely for that.
So no, Automator really is nothing like Hypercard. Hypercard's very little like anything except perhaps for Flash and Visual Basic. (If this was what you were saying or trying to get at, it wasn't readily apparent.)
Right, but look and feel doesn't necessarily equal "shiny" to everyone.
LINQ's order would be logical even if IDEs weren't even involved, since it leads you to define every variable before you use them, which is the default in every other place you use variables. If they had used SQL's order, this basic logical assumption would have been broken. (I'm not necessarily disagreeing with anything the parent said, I just felt this had to be pointed out.)
They didn't drop it out of spite as such. A lot of the existing VBA code was hand-rolled PowerPC ASM *and* based on a since-surpassed VBA version, in combination with an understaffed department for this sort of support. "Running to stand still" wouldn't really cover it, although I'm starting to wonder what has changed since; maybe it's just become more tenable to port the Windows version, or maybe people have been hired.
It's interesting to compare this to Microsoft's own research and actual usage data backing the entire redesign effort, which showed several real problems and which led the work.
I'm not all that in love with most other parts of Office 2007, especially not the underlying politics and company guiding the OOXML bullshit and anything-open-has-cooties thinking. But I have read up on how the redesign happened, point by point, and I can't fault them for not doing their homework.
It's a solid piece of engineering and craftsmanship (if you remove the horrendous branding like the "Office button"), but it's hard to judge the merits of the interface based on the first iteration of it, plagued by lack of customization and immense culture shock in anyone who sees it. The application of the interface to the programs might not have worked so well in practice as they thought it would in theory, but I think it's also fairly clear that "stay the course" would not have worked that well for that long.
There are plenty of beefs with the "crapware barrier" theory, although it sounds like reasoning Apple would apply. For one, the bucks prevent people who don't otherwise want to put out apps from making small, quick apps entirely for themselves, for personal use. Maybe Apple's just going to take their lumps or (inadvertently or not) nudge-nudge, wink-wink developers towards jailbreaking/pwning for those cases.
Furthermore, don't discount open source software (and I mean the open source construction process, not the Free Software licensing model and motive). To take one example, Adium on Mac OS X is one of the best citizens on that platform, away and beyond most commercial apps. You and Apple are right that this takes strong guidance, and per Sturgeon's law the vast majority aren't up to this. But the $99 fee that "they" can swing in the case of Firefox isn't per product or company, it's per *person*.
You're exactly right that it's not the absence of some crappy or unintuitive software that I'm going to miss on the iPhone; it's the presence of crappy or unintuitive software produced by people with money, and absence of potentially great software that could never be produced.
I'm expecting the community to hack this by creating distribution houses - something like software "labels" - that you can go through. But it doesn't solve simple stuff like beta testing without sending everyone your source code and having them download your SDK. I think Apple's greed and/or want for control (recall how old versions of APE triggered non-loading 10.5 installs six months back) has them overlook real problems.
The masses don't care one whit about the technology or licensing behind their apps, but the big deal with iPhone development is that to be able to test anything on a real phone "legitimately", you have to pay $99 for a developer license that's hitherto only available for US developers. For a "friggin $500 phone" (even if the 8 GB model is nowadays $399, so $400), this is *not* a big deal if you're planning on selling software, but a big deal if you're planning on releasing free software.
On no other mobile platform that I know of do you have to pay just to be able to try it out on a real device or just to be able to release freeware, and with the iPhone we're talking, in terms of user interface, the most technically advanced mobile platform. No wonder people are a bit upset that they can't use it. There's no accelerometer and no location awareness in the Simulator.
The "US only developers" thing is obviously to limit the test group to a manageable size until the SDK is really ready. We don't yet know what of the other restrictions are here to stay (I'm betting code signing and the App Store-only distribution are the most ingrained), but I really hope that it's a ruse and that most of the real brain farts will have been removed by WWDC, when the final SDK drops.
Yes, accounting reasons is probably why they did it, although not *really* because then they would have charged just a few bucks like with the N enabler.
If the update is to be seen as a dry run, I still think it's curious since most of this process is not at all how applications will be delivered with the SDK. (I know as little as anyone, but I have a very hard time imagining new firmwares for delivering third party software.) And my takeaway from this is that the update was not a dry run, or a dry run of a specific component still applicable.
I could perhaps imagine that people would get a Mac mini to develop iPhone applications, but it sounds absurd that Apple would have people who already own Macs buy new multitouch-equipped laptops. (And you can't even see the screen, obviously, on the trackpad.)
Signed applications means that the applications have a cryptographic signature attached to them with *any* trust root, not that they have to be attached to any specific trust root. There's a leap from "applications have to be signed" to "applications have to be signed by Apple" or the more likely "applications have to be signed by an authority whose certificate is trusted by Apple". But let's say that happens: unless all development happens inside a simulator in software (and good luck testing multi-touch then), I definitely think that there'll be a way to run "untrusted" apps, and this way will be exploited to run free apps. I think Apple knows this.
The iPod touch update was curious - the apps were already in the new firmware, and the update just "unlocked" them. (The update weighs in at 9 KB.) Since people won't get to download new firmware every time they get an app, this doesn't confirm much, although I agree that it was probably a dry run of some component in the whole scheme, most likely signing.
It's interesting that everyone takes for granted that "getting from iTunes and then syncing it over" will be the only way to get apps. It's likely to be one of the ways, but Apple has revealed nothing. It's all speculation so far.
They already host downloadable Dashboard widgets and provide links to all sorts of software on their site and host the world's biggest podcast directory at no fee for anyone, producers or users. I don't see how helping to host applications that could solve every non-hardware related aspect ("3G!") of their product would be *bad* for them, even if some of those applications were free.
I expect to see some way that Apple will help people sell their apps if they do end up with some sort of iTunes app store, but one approach doesn't rule out the other, especially since it likely won't be that easy to get access to their payment/transaction system.
The latest official word on this that I can recall says: "The main idea behind the Optimus Maximus key design is that the part with the display is fixed, while the transparent cap is moving, pressing a Cherry switch underneath[..]"
The displays are suspended below the actual switches and don't move when you depress the key.
Shouldn't, no. Imagine Apple courting carriers to implement stuff on their network so that they could deliver visual voicemail to a network that the phone isn't bound to. I'm not saying I wouldn't like ubiquitous visual voicemail, but it wasn't implemented before in that way and I don't think any carrier would take them up on the offer.
Fair point - unlimited data has been something I haven't seen offered myself unless you spring for a business plan or go for the full flat rate (unlimited everything) plan.
Couldn't agree more, I just hope their "multi-year exclusive" isn't that "multi".
Of course they can't. It is dumb. Your argument is correct.
In this specific instance, I happen to agree that there should be unlimited data plans, and there aren't that many that aren't also focused on "unlimited anything" (flat rate), at ridiculous prices.
There are three reasons I see for Apple ever wanting carrier exclusivity in the first place:
* Visual voicemail. New functionality on this order demands special implementation.
* Unlimited data. Regardless of 3G or EDGE, data on *any* cell phone that's not specifically a 3G modem tends towards ridiculous fees. If Apple had released an unlocked iPhone and asked for unlimited data plans, the carriers would laugh and ask if they also wanted a pony.
* To gain a foothold in the total clusterfuck that is the US mobile market.
They're all pretty sad reasons, which doesn't mean they aren't probable. And it just sucks that Apple took their strategy to countries where it's not really needed except to remain consistent with the AT&T agreement.
That makes some sense. They've lost the road warrior market if they don't have swappable batteries, but with LED backlighting, most things on the motherboard, a 1.8" drive (and SSD if you choose so, I suppose) and a minimum of extra ports (and optical drives) to power, they may be able to crank up the battery life to an acceptable level. Rumor has it that the advertised 5 hours were recorded with Wi-Fi on, as opposed to the standard "every feature off, RAM disk, screen minimum brightness, CPU napping and downclocked and fair wind; oh, and it never actually *did* anything during that time, this is just a theoretical maximum" industry fare. Then again, road warriors would probably like to see bigger hard drives.
Since the Air's battery spans the entire width of the computer, I don't think they could have.
Because when they don't need to make a hatch for the battery, they could make it thinner.
Seriously: the reason MacBook Air exists is because it's light, and because it's ridiculously thin. It's also interesting because of the trackpad and its multi-touch gestures, and because it has slightly different tradeoffs when compared to other subnotebooks which makes it a fairly unique product - MacBook Air has a faster CPU and normal-sized screen and keyboard, virtually every subnotebook has more features and ports in every other aspect, but piddly screens and keyboards.
The reason MacBook Air exists is because it's a unique proposition in the market (per the above), not because it's really convenient to service yourself or fast or extensible. There are other products for that.