20 odd years ago...I had been doing the usual round of physics graduate interviews, GCHQ's was a little different. After getting the security pass to get in and being escorted to the interview room, they told me that I wouldn't be able to ask any questions about the job (except pay). Or rather, that I could ask if I liked, but they weren't going to answer. Weird.

The point I guess, is that GCHQ don't recruit clandestinely like spooks, even if the interview process is odd. They're part of the civil service, they advertise in the paper, and recruit graduates in the milk round.

Cost of roads, parking spaces, *were* explicitly included in the study. I've seen estimates done for the air pollution side of things but it skews things massively, depending on what you estimate the health care costs due to pollution to be.

The article misrepresents the Ofcom report. Here's what the report actually said:

Source: http://stakeholders.ofcom.org.uk/market-data-research/telecoms-research/mobile-not-spots/mobile-coverage-for-consumers/

They go on to say that this may be in part due to the complexity of switching between 3G and 2G and that it can be mitigated by turning off 3G in your smartphone in rural areas...construing this as "users should invest in mobile phones different than latest Smartphones" is a bit of a leap.

If the reason for it being spherical is to allow it to recover being rolled on the ground, why not build the shell as a gömböc, so it'll always self-right?

I'm working on it. To get text out of pdf.js as is, you just implement a TextGraphics object (like their existing CanvasGraphics one) and just implement the text and coordinate transform commands. There's lots of ways of getting that into a copy/pasteable form afterwards, but its early days and I'm just coding up the OCR-ish algorithms needed to infer reading order from non-tagged pdf (the most common case).

I'm not associated with the project, but this is on their todo list too, and someone else might get it done before me. But it will be done.

Wrong. The test suite compares the canvas rendering against reference images (potentially generated on a different device).

Ok so I got bored and wrote a quine patent application. Almost as painful as COBOL. Havent tested it by running it through the lawyers.

METHOD FOR RECREATING THIS PATENT APPLICATION

CLAIMS

1. A method for constructing a textual representation of this patent
application, by executing the method described in its claims; the
method comprising:
(i) emitting the text result of applying a decoding procedure to a
data string S, followed the content of S, followed by a period
character ('.').

2. The method according to Claim 1, wherein the decoding procedure
comprises:
(i) splitting the data string S into pairs of characters H;
(ii) converting the pair of characters H into a number N via a primary
decimalization procedure;
(iii) Returning the result of replacing each number N by the by the
corresponding ASCII character.

3. The method according to claim 2, wherein the primary decimalization
procedure for a character pair H comprises:
(i) Splitting the character pair H into two characters, the first H1
and the second H2;
(ii) Converting the first character H1 into a number N1 via a
secondary decimalization procedure;
(ii) Converting the first character H1 into a number N2 via a
secondary decimalization procedure;
(iv) Returning the result of adding 16 times N1 to N2.

4. The method according to claim 3, wherein the secondary
decimalization procedure for a character C comprises:

(i) if the character C is a digit from 0 to 9, then the result is the
value of that digit;
(ii) otherwise, the result is 55 subtracted from
the ASCII value of the character C.
(iii) Return the result.

5. The method according to claim 4, wherein the data string S
comprises:
4D4554484F4420464F52205245...(and I had to remove the rest of
the hex-encoded version of the text above because the slashdot
lameness filter objected. Quite reasonably, I think)...
65696E20746865206461746120737472696E6720530D0A
636F6D7072697365733A0D0A.

The code in the patent application has been translated into legalese, wonder if that's done by machine? If so, it would produce predictable patterns and be compilable back into C.

My next entry for the Obfuscated C contest will be*:

A method or apparatus for compiling a patent into a C program that, when executed, outputs its patent.

Inventors: Bazzargh, W.V.V. Quine

*) of course I can't be arsed actually writing this

That's not the link to released betas. This is:

http://www.mozilla.com/en-US/firefox/all-beta.html

You'll notice FF4 isn't there. That's because the article has jumped the gun and is pointing you at a nightly instead, almost certainly not what you want.

As the weekly status meeting minutes say, the beta is coming soon and what is there right now is the nightly, for developers.

If we got hundreds of civil engineers together and told them about this, just think what we could power with their watering mouths!

Oh I've set them off again...

That's exactly the same horribleness that made me drop it. However you can actually get that feature to work: once the menu appears, lift your finger, NOW tap the menu. Of course if you move your finger AT ALL the menu is gone and you are left dragging over a bookmark you didn't want to open (it won't ever reappear again)

Its also inexplicably worse than having these bookmarks as icons on your phone. At least there you can drag them to reorder them - in opera mini, you have to re-bookmark something to have it appear somewhere else (or maybe there's a trick to this one too)

What purpose does it serve to skip version numbers, except for some political or media-relations reason?

Work was going on simultaneously on 3.7 and 4.0 branches of the code. There is an overhead in doing that, eg builds of both could be failing, who's looking into that, etc. Not least of your problems is getting developers who're working on shiny-new-stuff (4.0) to care about incremental-updates (3.7)

Version numbers are just marketing. The linux numbering system changed not that long ago, and every so often there is a bunfight over it (there was talk of making releases last year version 9.x, matching the year); and the discussions are always about how this would be perceived, since what matters to developers is just the git hash anyway.

advocated in a recent paper the 'cognitive infiltration' of groups that advocate 'conspiracy theories'[first link]... Sunstein has also recently advocated banning websites which post 'right-wing rumors' and bringing back the Fairness Doctrine.[second link]

What's interesting about these two bits of the summary is that both are based on the same article by Sunstein, summarized differently by different wingnut websites. I have to conclude that the submitter didn't read the article. At all. Hey, I can quote out of context as well as the next guy:

government may do best to ignore conspiracy theories and theorists even if it justifiably fears that they will have
harmful effects, because government action may make things worse.

Does that quote misrepresent the article? Yes, it does. But actually, less than the summary does, since Sunstein actually advocates 'ignore' in some circumstances. However - bans? If you read the article you'd see that banning is an option he explicitly rejects!! (clue: its option 1 of 5, where only 3, 4, 5 make the cut; its the only time the word 'ban' appears in the paper...). There might be something interesting to say about this article. But the stuff you've linked to in the summary, and the summary itself? That's just so far off target, its not even a decent starting point for a conversation.

Can we get back to the news for nerds, where someone claims Knuth advocates using O(n^2) algorithms, just because he mentions them (rolls eyes)

There's no such thing as an interpol agent. They delegate to national agencies (ie the DoJ) who do /not/ get immunity. What they do have is a bunch of committees and advisors, and a (shared) database of people 'of interest'.

Somebody's been watching the man from UNCLE a few too many times

Ok how about... a uuencoded[1], corrupt[2], powerpoint attachment inside an email that got pasted (as the raw text of an email) into a word doc, reformatted, and then mailed to me (as a base64'd attachment).

Got sent that ~12 years ago by a PHB who wanted help getting the powerpoint out of his mail. It wasn't that hard - I wrote a little uudecoder in perl that started dumping when it saw the magic bytes for OLE...that format has a LUT for 512-byte blocks of the doc at the start, if you have trailing junk its just ignored. I mailed this back to him, and hey presto, it worked. He then asked me how I did it. I wasnt about to explain this to someone who can't use the 'forward' button in their mailer so I just stripped the comments and spaces out of the perl and sent that back.

-Baz

[1] the microsoft flavour. this was back in the day when there were multiple uuencode alphabets.
[2] extra bytes at start and end.