That strip shows polling behavior, not interrupt driven. Interrupts are my preferred method of dealing with the world. How much time have you wasted checking email or Slashdot when there was nothing new? I know the figure is pretty high for me... That said, context switches are indeed expensive.

What I find works best for me is an unobtrusive notification, that I can address when I've wrapped up my current train of thought - the CS analogue would be a top-half/bottom-half split in my interrupt handler. The top half notices that my phone has beeped, and I schedule it for later.

A good example of why windows shouldn't steal focus, but rather irrelevant to the subject at hand...

Actually, it's a little known fact that 1996 was only 6 days long.

To some degree you attack a straw-man here. The bewilderment expressed in these comments is at the apparent waste of resources - setting up a means of informing security when the burner is used is significantly harder than simply removing/disabling the burner, and no easier than selectively enabling it. I question the notion that burning CDs may have been a regular part of business - they surely don't quarantine a section of the office in response to routine behavior. It is precisely the fact that they went to such lengths and seem to care so much and yet aren't taking what seem to be more reasonable precautions that leaves us confused.

I don't think it would. Consider the last sentence of the passage you quoted.

This is manifestly not the case when I am given a modified version of GPLed code and told I cannot distribute it.

Note specifically that in the examples above, it's a question of what someone agrees to do with code he has written. In our example, it's a question of restricting what someone agrees to do with GPLed code they were provided. If - a big if - the employees are taken to be recipients of the software, the employees are either implicitly licensed the software under the GPL or the company is in violation of copyright.

Actually, I think it's only partially incorrect.

Point 6 of the GPL:

The question is whether an employee asked to use the software can be considered a "recipient" of the software. For this, I can see arguments both ways - and I do not believe it's been tested in court.

I've been saying for a while now that precedent should be taken as bug reports - if there is room for interpretation, the original law was unclear, and should be amended.

Liability should fall with whoever deploys the software - they are the only ones in a position to know of the potential downsides. Contract can shift liability after that point, and insurance can get involved at any point, but that's where it should start.

For FOSS, this means that you're using it at your own risk, unless you're paying someone to take that risk on for you (a business opportunity in its own right). This seems appropriate.

Trademark exists to prevent fraud and protect reputation. Unlike data (protected by copyright, patent, and trade secret), reputation cannot be trivially reproduced - the value of a mark derives from its ability to inform, which evaporates if there are no restrictions on what the mark can apply to. While flaws in the current system are sometimes abused, and there is unquestionably room for improvement, the existence of trademark does not geometrically decrease the value in the system (as the other forms of "IP" do). I suspect this is why there is less railing against trademark than against copyright or patent - it's simply *less* broken.

Ten minute walk through the woods. Win.