Apple's iOS App Store rules state that all in-app purchases must use the In-app purchase mechanism. In addition, you cannot link to an external purchase mechanism. Finally, Apple charges their standard 30% fee for purchases for in-app purchasing.

The issue in the Dropbox SDK is that it fires up a web view for authentication, and the page it went to was not properly sandboxed. By creating an account, then clicking a link to go to the desktop version, you got to a page that let you pay to upgrade to a pro version. At this point, it broke the rule above. Since this was functionality within the SDK, all the applications being submitted to the store using this version of the SDK are breaking this rule and getting rejected.

My guess is Dropbox released a new SDK version that sends the user to a different web page for authentication, this one being properly sandboxed. I do not know what Dropbox would be talking to Apple about; unless Dropbox API support is added to the OS, you would need the Dropbox client installed in order to be able to even support in-app purchasing of a pro account.

I believe (like FreeBSD) they are fine with GPL v2. It seems like projects switching to GPL v3 is motivating Apple and a few other of the BSD'ers to create new open source projects for all the functionality they were getting from these projects. I love that it finally motivated someone to invest heavily in coming out with a gcc alternative (clang + llvm).

I have noticed a pattern that apple's replacements only implement the newest defined functionality or protocol, however. For an example, there is work going on to replace libstdc++, but the replacement is only meant to target compilers that support C++11. So I wouldn't be surprised to find out that say, a Lion Mac has trouble connecting to a Windows 2000 server using their samba replacement.

I've been wondering for a now if there is a project somewhere to come out with a BSD-licensed replacement for bash.

OS X no longer ships Samba as of 10.7 (due to the switch upstream to GPLv3). So yes, if by broken you mean 'was replaced with an entirely new project', it will remain so.

I believe all of the large companies had rights to the patent as part of the Intellectual Ventures portfolio, and part of the deal of the sale of the patent to Lodsys included that companies who had license to the portfolio retained the right to use the patent.

I believe apple has a head-start on the technology, not an exclusive. Intel's stock parts do not include thunderbolt, stock video cards do not support thunderbolt and I believe most PCs have HDMI or at best Displayport, not Mini-DP.

Until PC vendors want thunderbolt enough to get it integrated into third party video cards, or until Intel ships Ivy Bridge which is supposed to include Thunderbolt for its integrated video, Apple will be the only vendor producing computers with Thunderbolt.

Even then, it looks like Sony is interested enough in the technology to try to create a proprietary version of Thunderbolt which uses additional connects on a USB port, and most likely does not support the DisplayPort video channel.

Firewire 800 is slower, what about firewire 3200?

Damn it, uid envy

Except those do not have access to the google marketplace.

People seem to think that this was done for piracy, or done by extraordinarily clever hackers through a lot of time and pain.

Thats all bunk. The whole reason people hack these master keys is to sell a butt-load of t-shirts.

Your assumption may be that the java code to handle asynchronous IO is optimal. However, there may be a big performance hit not from the underlying system, but from the particular JVM’s implementation of asynchronous I/O.

I know in particular that the NIO system on windows used to use evented sockets rather than IOCP, which among other things meant that IO events were shoved into the windows message pump, and only 32/64 sockets could be handled at one time.