Slashdot Mirror

Look kids. Get over the small minded philosophical hangups. Understand that the MacBook OS is a BSD kernel + the GNU OS (tool chain) + Plus the NSstuff that Next brought. That's it. The vast majority of code is already open, because it has been developed by the community over 30 years.

the XNU kernel is an evolved version of the XNU kernel from NextSTEP that uses some BSD components, CMU Mach microkernel components and C++ I/O Kit which replaced NextSTEP's ObjC DriverKit. It's not a "BSD kernel" per say. the toolchain is definitely not GNU at all. LibSystem uses no GNU code at all. It uses the BSD standard library libc, not glibc. clang is the compiler, not gcc as that's something they got rid of many years ago. They do still use some software preinstalled that are under GPL but it's no "toolchain". See Apple’s great GPL purge.

And none of this is a bad thing. It is all perfectly logical, and useful decisions. glibc is horrible bloatware, gcc has a license problem (go read about how it closes off, intentionally, plugins for things like static analysis).

It works, and works well... but I don't like how they have wrapped LLVM/clang in Xcode. I think it's just stupid that POSIX programs don't always work the way you expect or are missing (e.g. umount -> diskutil unmount ) but for the most part, it's UNIX that Works.

A number of important components are completely closed which are needed to boot XNU on its own, like PlatformExpert. So you're not exactly correct in your statement here.

Yes. They think they need something for hardware vendor lock in. They are right about that, given their business model (and no, I don't have a solution to that).

Look kids. Get over the small minded philosophical hangups. Understand that the MacBook OS is a BSD kernel + the GNU OS (tool chain) + Plus the NSstuff that Next brought. That's it. The vast majority of code is already open, because it has been developed by the community over 30 years.

the XNU kernel is an evolved version of the XNU kernel from NextSTEP that uses some BSD components, CMU Mach microkernel components and C++ I/O Kit which replaced NextSTEP's ObjC DriverKit. It's not a "BSD kernel" per say. the toolchain is definitely not GNU at all. LibSystem uses no GNU code at all. It uses the BSD standard library libc, not glibc. clang is the compiler, not gcc as that's something they got rid of many years ago. They do still use some software preinstalled that are under GPL but it's no "toolchain". See Apple’s great GPL purge.

A number of important components are completely closed which are needed to boot XNU on its own, like PlatformExpert. So you're not exactly correct in your statement here.

And depending on how it goes about it, I may have no problem with that.

There was no GPL purge.

Thanks, Baghdad Bob, but I've been monitoring the purge for some time.

You appear to believe that argument from authority trumps sense. You have it backwards.

First off, I was merely pointing out that GP's statements were factually accurate, and therefore did not deserve ridicule. I said you have the choice to ignore those recommendations by standards organizations if you want. Go ahead. But if you want to argue from a system of logic rather than authority, you may want to reconsider your methods.

Secondly, you have it backwards and believe in some mythical authority that never was consistent. If you actually were correct about 1024 being used in ALL cases of binary standards, you might at least have a leg to stand on. But that's simply not the case -- the system is inconsistently applied and has been from the start.

To put it another way, even if you got every committee of busy-bodies on the planet to agree that 2+2=3, they and you would still be wrong.

Not if I were running a "three-legged race"! You see: even 2+2 depends on context! See what I did there? That's your logic in a nutshell. But the situations that 2+2=3 are relatively rare and hard to define as a specific class, so we all agree to be CONSISTENT and accept that 2+2=4 as a DEFINITION, and regard the "three-legged race" as a different thing, not something that trumps the standard definition of 2+2=4. Binary prefixes are needed enough that we need a system to reference them, and one has been defined. Just about everyone in the "biz" these days is aware of it. What's your excuse?

Are you seriously claiming that if Apple were in Microsoft's position things would've been better? Imagine no OEMs, no Dell, no Lenono, no HP and having to buy only from Apple and paying both a huge hardware and a software tax. Imagine paying Apple 30% tax for every PC app ever sold, if it were ever allowed to be sold by Apple ie. Remember how they say on Google Voice for a year before rejecting it?. If anything, Apple's lockdown is far far worse than even MS could've dreamed of. Remember how they killed off Psystar and how litigious they are?

Apple's already stopped shipping any GPLv3 apps. http://meta.ath0.com/2012/02/05/apples-great-gpl-purge/
http://apple.slashdot.org/story/11/03/24/1546205/apple-remove-samba-from-os-x-107-because-of-gplv3

Just because Apple is able to keep their version of Halloween documents from the prying eyes of the public doesn't mean that they don't actively seek to do the same thing or worse, and it seems to be pretty naive to think otherwise if you go by their actions.

Apple Airport Extreme. Well, actually a Time Capsule, which is basically an Airport Extreme with a big hard disk. I've been IPv6-ready for over two years now. Here's what I wrote about the process, with screenshots at the bottom.

Sure. I've written a post on my web site with addresses for the first round of complaint letters. We've got a few months to try and scupper the deal. Spread the word and get writing.

What mystifies me is that Apple would store the time internally as UTC instead of going pure local time.

Because date and time calculations are a bit more complicated than you think. In general there is no right answer to the problem of time zones, other than making them explicit on every appointment. (See "Automatically adjusting dates and times in a calendar is hard" in my linked article.)

Yup, I predicted this is how it would go down too.

Said page says

2013: Mac OS X 10.9 no longer runs unsigned code. For that, you need to buy Mac OS X Professional, or the developer SDK and a signing key.

and

The Mac is now locked up the same way as the iPhone or iPad.

So where can I buy iOS Professional? For Mac OS X to be locked up the same way as iOS, there would have to be no OS from Apple, running on iOS devices, that lets you run unsigned code without the developer SDK and a signing key. Requiring Mac OS X Professional for that isn't as bad if you can run OS X Professional on any machine that can run OS X End-User - if OS X Professional is available to all, the pain would be 1) financial, if OS X Professional is more expensive and 2) time and convenience, if you have to install OS X Professional on top of OS X End-User on machines that ship with the latter.

(And, yes, I think it'd be cool if Apple offered, for iOS, a switch you could flip to allow running non-App Store approved code, even if doing so also flips the "your machine is under warranty" switch to "off".)

Yup, I predicted this is how it would go down too.

Getting everyone hooked on the app store is phase 1. If it catches on fast enough, they may be able to start imposing the lockdown in OS X Lion--that's why they're launching the app store now. Otherwise, they'll wait until the release after Lion.

After 23 years as a Mac user, my days as an Apple customer might be numbered.

Extracts from George Orwell's "iPhone Four".

Beat you to it...

Imagine how sick of it I am, since I posted the same idea before xkcd, and now get people sending it to me as a wonderfully amusing thing xkcd thought up...

I like so many others, have tried, again and again, politely and impolitely, to get FF to focus on so many problems... Like the bookmarks editor...just hopeless Like the loss of control of privacy functions...

Yup. I tried to convince them that per-site script and cookie permissions need to be part of the core browser, and not require add-ons which increase bloat and cause compatibility problems... but no, they're too busy building useless address books that nobody wants, and trying to ram Ogg Theora down people's throats.

So when Chrome got per-site script and cookie controls, I switched.

Chrome has all the functionality of NoScript and ABP built in with a significantly lower memory footprint?

Here's what I wrote about it. You still need an adblocker, but cookie and script permissions are built-in.

If it could add the ability to manage the exception list while you are looking at the page (without diving into menus) the way NoScript does, then I would switch to chrome in a heartbeat.

It already has that. Small icons appear to the right of the entry bar when cookies or scripts are blocked. You can click the icon to turn on scripts or cookies for the domain.

More info and screenshots at my web site.

People are laughing at me when I suggest that future iMacs will have app store lockdowns and now will be "ad-supported" to boot.

No, it's my big fear too.

You are assuming that the entire purpose of DLC is not to obliterate the secondary market for games. There's no reason to release DLC except to ruin the used game market.

Not really true. It's also a great way to jack up the price of games without it being obvious on the sticker.

For example, you buy the latest online FPS, and find that there's a map pack ($15) and a weapons pack ($15) available as DLC. You have to buy them, or you can't find anyone else to play multiplayer with. Hey presto, the $60 sticker price is actually $90.

Then after a few months when you've moved on to the next game, you go to sell your game used. Except the Game Of The Year edition is now out, with the DLC included, for $30. Which means your used copy without any DLC is worth practically nothing, rather than $25.

I noticed this trend a while back. Ultimately, I think the game publishers are playing a very dangerous game jacking up prices this way, because there are people like me who would have paid $50-60 for a new game, but will instead wait and buy the cheap copy a year later rather than be doubly screwed over by DLC.

It's possible that the extra money they make from suckers who still buy games brand new will more than offset what they lose from people like me no longer buying anything at first release, but I'm doubtful.

Programming is becoming nothing more than cutting and pasting, especially with languages like java, that provide libraries that do "the hard stuff" and let programmers concentrate on "programming".

You say that, but take JDBC as an example. All of Sun's examples are wrong and can leak resources. All their reference code is wrong too. O'Reilly's onJava has incorrect code examples. Sun even has an article about resource leakage and the importance of getting JDBC calls right--and the article gets it wrong. I've written about this on my web site.

It's not just JDBC, there are lots of Java APIs that are routinely used incorrectly. Ever seen a Swing application UI freeze on you? They probably got the threading wrong in the initialization.

The fact that incorrect code is so prevalent points at cargo cult cut-and-paste programming being the norm. Use of standard libraries really has nothing to do with it, as cut-and-paste use of libraries is as bad as cut-and-paste anything else. This, in turn, is why Java has a reputation for "memory leaks".

Yeah. I'm a Mac user, and I'm convinced that Apple is strangling the iPhone. I'm waiting until the 27th to see if they launch a non-jailed device. If they don't, I'm giving up and switching to Android.

Apple's partnerships with Microsoft have always ended badly.

Everyone's partnerships with Microsoft have ended badly. I don't think any company has done business with Microsoft without getting shafted. Vivo, Real, Kodak, IBM, the list is lengthy.

So, for Apple's sake, I'd think them making their own search engines/tools or sticking with Google, may be their best bet.

I think un-jailing the iPhone would be their best bet. That's what I think is really strangling the platform right now, causing developers to give up and users to lose interest and consider switching to Android. There's no way a jailed, intentionally crippled platform from a single vendor can compete in the long term with a multi-vendor open platform, even assuming feature parity (i.e. even assuming Apple came up with alternatives to Google's search, mapping, etc.)

Really, it's like 1985 all over again, but even worse, because at least the Mac was open and you could run what software you liked on it. Imagine how much worse the Mac would have fared if you had to get Apple's approval for your software to run on the Mac, and you could only buy Mac software from Apple.

But if this new device blurs the line too far away from 'throwaway gadget' to 'computer' Apple may run into trouble.

Yup. I'm interested in an Apple tablet if it's a general purpose computer. I'm not interested in a locked-down and intentionally crippled giant iPod. Apple's grip on the iPhone is killing it, and I see the tablet as a critical test to see whether Apple understands that or not.

My web site ATH0.com has an old modem ad on the front page, I scanned it from an 80s magazine.

Without the producers being able to subsidize niche channels through fees for their popular, flagship channels - which is, of course, exactly why they sell channels in packages like they do now - the price of those niche channels will go up dramatically for those who choose to subscribe to them.

I think you're wrong. The truly niche channels are currently usually sold only as add-ons or in premium packages, precisely because they don't have the clout to force their inclusion in the core packages. For example, if I wanted the Science Channel or Logo via DirecTV, I'd have to upgrade to the $61/month package. In the mean time, I was left subsidizing channels like ESPN and FOX News, which frankly don't need subsidy.

[I've written about the whole a la carte thing in more detail.]

According to this, EarthLink over Time Warner Cable's pipes will also be capped:

EarthLink high speed cable provides service in Austin. As part of the regulatory compromise which allowed Time Warner to purchase AOL, Time Warner cable are required to allow EarthLink to make use of their cable. However, Time Warner have claimed that they will be capping the Internet service of EarthLink customers, and as yet nobody credible from EarthLink has denied this.

First off, Palm don't own PalmOS. It's owned by Access, who bought PalmSource.

Secondly, PalmOS's plug was pulled back in 2005, when Access announced no further development work would be done on it.

Thirdly, Palm didn't *decide* to pull the plug; their license from Access to ship new PalmOS devices expired, so they have no choice.

I wrote about all this back in 2005 when the news went around. I guess everyone's forgotten.

You're half there.

Access owned PalmOS, and in fact PalmOS was killed in late 2005 when Access ceased development and moved to the Linux-based ALP (Access Linux Platform).

This announcement is actually just Palm admitting that they can't afford to release any more hardware that uses an OS that's been dead for nearly 4 years.

No, it's not, because hard disks sizes have NOT been around since before than the most common operating systems and software environments (such as DOS etc. popping up and calculating disk size divided by 1024)

Wrong. Hard disks have been around for a lot longer than DOS. And they used decimal multipliers to measure capacity way back then.

When your disk is "200GB" and every OS written before 2 years ago says "195GB", and this cannot be put down to the old adage of "well, formatted capacity is less than disk capacity" bullshit excuse (not even the worst filesystem ever would drop that much of of the disk space) or anything else but boldfaced cheek.

Who is putting it down to "formatted capacity? Everybody understands that it is because of the two different systems of units. So, where's the cheek in using units correctly?

Of course your view of it is entirely revisionist, in that somehow a special new naming methodology that hadn't been standardized is NOW used which exonerates them for all the mislabeling in the past.

But there was never any mislabeling in the past. You are the one playing revisionist. It has always been the binary system of units that was the odd one out. It is ONLY used in programming and measuring computer memory. It was never a standard. Everywhere else in the science/engineering universe, "mega" means "1 million" and "kilo" means "1 thousand. It was the programmers who screwed it up by being lazy with terminology, not the hard drive manufacturers.

Hard disk manufacturers were expected to specify the disk size in terms of how you will see this size in the applications you're meant to use it, if they are marketing to people who are using these applications.

Say what? I expect them to make storage devices. They accurately specified the size of those storage devices, so I don't see what was done wrong. The hard drive manufacturers don't control the OS developers, so how is it their fault that their product was being misrepresented by software?

It makes no sense to measure hard drives in binary units.

but the hard disk manufacturers were sued, and settled exactly because they could not hide behind "engineering" to explain why the advertised values differed, and they could not explain WHY they did not clarify it on the packaging or specification sheets of the products.

Oh, bullshit. One manufacturer settled without admitting wrongdoing. It's often easier and cheaper to settle a lawsuit than to argue it. And it's not about hiding anything. It's about the lack of technical competence in most court rooms. It's highly unlikely that they'd get a fair trial, as most courts wouldn't understand the technical issues and subtleties.

They just kept the biggest number and got caught out. Hence, big fat lie.

Again, what do you mean, "caught out"? They were caught doing something which they'd been doing for decades, which was never a secret, and never intended to mislead?

You are accusing people of lying, and you have no evidence whatsoever. Is that something you normally do? It's pretty disgusting. Nowhere in your rants have you provided one shred of evidence. You are just spinning a story that matches your pre-made conclusion.

Look anywhere, and you'll see that a byte being 1024 bits has never been "standard", and until recently it has not been standard practice to clarify the meaning in boilerplate text. The situation has been a mess for a long time, and to simplistically blame it all on hard drive manufacturers is absurd. Especially as, if anybody started this mess, it was programmers abusing the SI prefixes in ways they never had been in any other field. The hard drive manufacturers were at least sticking with a more logical and consistent system.

Some more sanity on this topic.

So is there some easy way to get rid of the useless new KDE 4 desktop and replace it with a folder view that works like every other GUI? Like you can rip off the useless new launcher and replace it with a hierarchical one?

Representing times and dates as "seconds since epoch" is itself fundamentally broken.

ODF should use a representation that actually allows future dates to be represented with second accuracy.

Yeah, I had exactly the same idea over 3 years ago. It doesn't even need to be the government that does it.

KickOff's functionality may have been a big improvement for you, but for some of us it is a huge leap backwards.

I hate to harp on about it, but that start menu has got to be fixed.

I see that it's now resizeable, but that doesn't fix the basic problem of it hiding all the contextual information about where you are in the menu structure and being useless for people who rely on spacial memory.

I'm seriously wondering whether I'll have to switch to GNOME or Xfce.

I posted the exact same joke to my own site before the xkcd comic...

No, you're wrong. Kilo has never consistently been used to mean 1024 in computing. The same is true of mega- and giga-. Your 2 gigahertz CPU is not running at 2*1024*1024*1024 hertz.

The new XP-style kicker replacement is an absolute abomination to use. Too many clicks, practically impossible to browse the program hierarchies quickly.

Yeah, you're right.. I wasn't trying to make a distinction between the powers of 2 and powers of 10 in that post, simply to correct a roughly 1000-fold mistake in a parent post.

I've grown up thinking 'kilobyte = 1024 bytes, megabyte = 1024 kilobytes', etc, FWIW.

Perhaps "kilo byte" should mean 1000 bytes and "kilobyte" should mean 1024 bytes. It depends on whether kilo is being used to mean the SI prefix of 1000... insanity.

An interesting post on the whole mess: http://meta.ath0.com/2005/02/23/a-plea-for-sanity/

No, the 6.x to 7.x X11 upgrade was a major problem for ATI FireGL T2 users. I know, because the problem bit me too in Debian.

(like who a hard drive isnt EXACTLY the size thats claimed)

The hard drive thing is different, though. Hard drives are exactly the size claimed. The problem there was that the people complaining were using non-standard definitions of the units in question for historical reasons.

I think one opportunity that Debian continuously fails to see is to make very clear that Testing is always uptodate and always usable.

It isn't, though.

PAM in Testing was broken for months, and X in testing was broken for a while after the changeover to X.org. That's what led me to give up on Debian: 'stable' was too out-of-date, 'testing' was too unstable. By cherry-picking from 'testing', Ubuntu seems to be able to find a happy medium.

Yeah, CookieSafe makes cookies work the way they should do. It ought to be standard.

Damn straight. My summary of the story is: "RPM sucks so bad, and RedHat's packaging discipline is so bad, that even Eric Raymond can't stand it any longer".

I've been saying it for years. I'm amazed that RedHat don't consider RPM important enough to fix.

(My most recent case of RPM crapping all over its own databases and totally hosing the entire system irrecoverably was last October, so don't anyone try and tell me it's fixed. No, --rebuilddb didn't fix it.)

Any program keeping its logs in local time should be shot.

Like Linux and every major Unix, you mean?

I decided to do a statistical analysis of my music collection. My statistics dispute the idea that music got lost in the 1990s. For me at least, it was all up until 1994 and hasn't dropped much since then.

http://meta.ath0.com/articles/2006/10/27/proof-tha t-the-70s-sucked

No, Java is often unnecessarily complex. Most languages get by with arrays; Java has arrays, Arrays, Vectors, and ArrayLists, all with subtly different APIs. Ditto Hashtable and HashMap. Mostly this explosion of APIs has happened because Sun hasn't thought through the design before adding stuff to the language.

Then there's the whole EJB and EJB QL fiasco. Massive amounts of additional complexity added for zero gain.

Right. The problem isn't your employer using your SSN to identify who you are uniquely. The problem is dumbass companies that pretend that knowledge of your SSN proves you are that person.

I've written before that there's actually a free market solution to the problem. What it needs is for some well-funded activists (Gilmore?) to put together a nice big database of SSN info. We know all that info is available to any company that wants it.

Then, public announcements are prominently made in the press (NYT ads, paper mail notifications to every major bank and so on) stating that on 2008-01-01, the entire database will be made public for search purposes on the Internet. On that day, you'll be able to look up and verify anyone's SSN for free. That's the way it should be, after all--it's an identification number, not a password, and anyone can look it up for $20 from one of the many online services. We're just going to change the price.

This means that any organization currently using SSN as a secret identifier basically has to stop doing so, or face massive fraud and consequent liability lawsuits.

Yes, the issue of companies using trademarks to lock up GPL software is one I think ought to be addressed in GPL v3. I wrote to RMS about it, and also posted my thoughts. I think it's a genuine risk.

RMS himself doesn't have a problem with what Mozilla are doing, but I think that loophole is going to get widened if Mozilla get away with it.

Assuming the Wii really does run Linux, they will doubtless be using the TiVo hole to get around the GPL v2.

That is, they'll provide the source code with their proprietary modifications for the Wii hardware, but it'll be totally useless as the Wii hardware will be designed so that it will only run code signed by Nintendo. So the modified code will be useless to Wii owners, and also useless to everyone else as PC hardware won't have any use for the Wii hardware support.

And Linus will no doubt say that this is just peachy.

I think it's exactly the kind of crap the GPL was supposed to stop. If I purchase hardware and software that's GPL licensed, I should be able to modify the software and run the modified version on the only hardware it's useful for, the hardware I own. That's why I support RMS's efforts with GPL v3 and think they're a good thing. In fact, I think they should go further.

In fact, the use of trademark law to encumber GPL software and make it non-redistributable is something I've been predicting for a while; I recently wrote about it and sent a note to RMS suggesting that protective measures be considered for GPL v3.

RMS replied that he doesn't have a problem with what the Mozilla folks have done, but that there could clearly be a problem down the line when the loophole gets widened.