dbaron

Multi-process is happening just fine with Gecko, XPCOM, and XUL. It's the extension ecosystem (the do-whatever-with-the-internals nature of the extensions API, and getting all the extensions to upgrade) that's the obstacle. Nightly is currently multi-process (although not large numbers of processes, but there's a pref to flip); it just hasn't shipped in the release channel yet.

So we (Mozilla) are following the spec, which says what it does as the result of discussion/negotiation over a period of at least 4 years (roughly 2007-2011) involving browser makers and font foundries, both in terms of what browsers were willing to implement (not wanting something DRM-like to protect fonts) but what would lead to more font foundries being willing to license commercial fonts for use on the Web. So following the spec here isn't about blindly honoring some piece of paper; it's about honoring the result of a negotiation process that we participated in. WebKit unilaterally ignored this agreement, though it sounds like Blink may well change to honoring it.

Many (but not all) of us believe that same-origin by default is also the right thing for security, that it should be the default for new types of resource linking on the Web, and that not having same origin restrictions for things like images was a pretty serious mistake that we're still paying the security costs of (for example, with canvas image tainting).

My understanding (and while I've talked to people who are in the working group, I'm not in the working group or following the discussions closely) is that this was already (prior to the spec being modified) the common understanding of the people within the working group. It just hadn't previously been written in the spec, perhaps because it was thought to be obvious.

(And, given the context of what the group is trying to do -- build a do-not-track mechanism that advertisers are willing to honor -- I agree that it's obvious.)

Now that I'm home: the scan is from page 508; the relevant bits are from the end of page 504 through page 508, or perhaps all of Chapter 67 (502-509).

Personally, I'm stumped by the "without anything passing in writing" part, which seems to me to make the problem unnecessarily harder, since Pandarus was carrying codes for them printed on silk (to be sewn into the lining of clothing).

And... since I haven't mentioned it already: this is a great book. It's gripping, and reads a little bit like a movie for a good reason -- its author (who is telling the story of his experiences working on cryptography during the war in his early 20s) went on to become a screenwriter.

Having read the book in the past (though I don't have it in front of me right now), I think what it means is that he has to give Menaleus a way to identify himself in encrypted radio communications, such that if Menaleus is captured by the Germans and ordered to send certain messages (to misinform the British), he can omit his identity check without detection by the Germans so the British will know that he's transmitting under coercion. Doing this in a way that works if the Germans have decrypted his back traffic is hard -- I think the assumption was that they hadn't done so due to lack of resources; they'd instead torture the agent to get him/her to reveal the identity check.

Even worse, the buttons hosted on wd.sharethis.com set frequent timeouts that use up a measurable amount CPU (and laptop battery life).

(This is particularly bad when I open a large number of articles in tabs, to read later.)

Interesting to see this on HN a little over 2 years after I wrote it.

My main point wasn't so much about the magnitude of the taxes, but about the shape of the graph. Most of what gives this graph its bizarre shape is a result of the federal laws, not state ones; I just happened to be computing for a California resident because that's where I live, so I had the tax forms in front of me. (And it's also the largest state by population.)

(I'd also note that state-to-state comparisons of taxes aren't really fair without adding local taxes too, since different states have different relationships between the amount of money spent by the state government and local governments. In California, specifically, it's very hard to raise local (property) taxes, so much of education funding comes from the state government. In many states, education would be funded much more by local taxes.)

This is more the architecture that the W3C TAG would like the Web to have rather than the one that it has in reality. (There are substantial differences of the "no piece of software would actually do it that way because it wouldn't be able to handle the content on the Web" sort.)