The Sunday Times yesterday ran a story on Cambridge "night climbing" - the idiotic pursuit of climbing beautiful old university buildings and hanging your pants off the top. I'm a long-term advocate of this sort of silliness, so I was asked to come down to Cambridge a month ago and climb a few things with photographers in tow: the result was an article with a bunch of excellent photos (mostly not of me, thank goodness). I also got a few quotes in, and a plug for my Night Climbers of Cambridge online book - no URL, since apparently the ST is in state of denial about the existence of the Web.
Since there aren't any photos in the online version of that article, here are some for your entertainment - thanks to Jan Stradtmann for letting me use them:
These photos are (I think) copyright Jan Stradtmann - contact him if you want to reprint or generally reuse them.
A couple of years ago, I taught myself some rudimentary PostScript programming from Bill Casselman's brilliant Mathematical Illustrations book. PostScript is excellent fun, and I thoroughly recommend it to all weirdos with an interest in stack-based graphics programming.
Recently, while fiddling with MetaPost (another perverse but entertaining graphics language), I remembered using Python and PostScript to create a nice Bezier curve approximation to a cycloid. This was before I discovered PyX, which does that sort of thing for you, and it's a very illustrative exercise - but my not-quite-right first attempt actually produced a fascinating and very pretty result:
While the code that produced it is long gone, I'm happy to still have the result. Only mathematical illustrations can produce such satisfying errors... and just imagine what someone talented and artistic could do with the right mathematical/programming knowledge.
It's wonderful how small things can remind you of distinctive times in your past. A couple of weeks ago, while down in London, I was lying on my hotel bed when The Field Mice's "If You Need Someone" started playing on my laptop. It triggered a set of memories all at once and the whole feel of 8 years ago when I first heard that song. I was in my second year at university and among my friends I was the one who'd first gone and spent my 15 on the completely untested Field Mice best-of. We'd missed their heyday, but tweeness was of the zeitgeist as evidenced by Belle and Sebastian, and hell, best-of's have their own mystique in the right situation. My memories of that time are so distinctive - a second, more confident year in college, nights at the Q Club, discovering My Bloody Valentine and Stereolab via the college CD library, sleeping in, "getting" quantum mechanics, reading Snow Crash, playing Goldeneye, doing the radio show... It was a golden time for me, and in the traditionally cliched way, that's something I've only appreciated via the magic of the retrospectroscope.
I'd be delusional if I claimed to be an expert on current affairs, but surely I'm not the only one who finds the current British political obsession with "international terrorism" a bit odd. Today's main news is that Gordon Brown is determined to introduce more anti-terrorism laws when he becomes Prime Minister - the usual fun laws like arbitrary detention without trial, use of phone tapping evidence in court etc. etc., with a good dollop of haziness in case anyone tries to hold him to anything specific later. As far as I can remember, the last and only terrorist attack on the UK since 9/11 was the 7/7 London bombings. That's it. A variety of dodgy characters have plotted two-bit schemes that came to nothing, and well done to the security services for stopping them... but that's it. Now, I'm not so keen to play the I-come-from-Belfast terrorism superiority game, but for me one attack doesn't really justify 6 years of anti- terrorism obsession and Muslim-baiting. Especially not when much of said terror has been self-inflicted and our authorities have proven stunningly bad at actually making Islamic communities feel more integrated into British communities - in that respect we're really a long way behind where we were 10 years ago. (I should clarify that I'm so naive as to think that there isn't a genuine problem with Islamic extremism and its attitude to Western society - it'll take more than a few lefty-liberal-intellectual "hug a Mahmoud-ie" (sorry) blandishments to sort that out.)
I finally got pissed off that I couldn't run Java applications or Ruby Gems on my rented virtual server, because of
memory limitations - 80 Mb just isn't enough, especially when
gem is as boneheaded as it is about reading in the
whole gem index at once. So,
despite having had such good experience with Bytemark, I decided that I can't afford to
upgrade to their next best server package, so I've migrated to Slicehost. Just as I had
bought my 256 Mb "slice", Bytemark announced that they would be upgrading my package to 150 Mb and 10 Gb of storage.
Well, I'm sad to ditch them, but Slicehost is still offering an extra 100 Mb of memory at a slightly lower cost... no
contest. Plus, my server had got into the Linux "too many packages installed just for the hell of it" phase and it's
almost easier to move between providers than to attempt to juggle my files and configuration around on the same VM
while I reinstall it.
Anyway, my first impressions of Slicehost are very good: they know enough Web development kung fu to have provided very nifty, slightly AJAXy Web console apps and a nice slice monitor/DNS manager etc. And my slice was ready to use within 2 minutes of handing over my credit card details - that speaks volumes to me about knowing what you're doing. I've now upgraded the site to Radiant 0.6 and have a few random articles in the pipeline - it's nice to get out of the doldrums of worrying about which blog/CMS/whatever engine to use and to just get on with incoherently venting my spleen again :-)
While hacking together a generator cut combining system for Rivet I just got an enormously unhelpful error from g++, namely
from Projection.cc:3: ../../include/Rivet/Cut.hh:76: error: `map<string, Rivet::Cuts::BinaryCut>' specified as declarator-id ../../include/Rivet/Cut.hh:76: error: two or more data types in declaration of `map<string, Rivet::Cuts::BinaryCut>'
Fortunately I've seen this one before, though I couldn't remember at first. It turns out that that's C++ compiler error language for "you forgot the stupid semi-colon on the end of your inner class declaration". Another reason to relegate C++ to the garbage bin of history! Incidentally, I just discovered that Bjarn Stroustroup used to be at my college in Cambridge... hopefully I'll run into him at a dinner sometime and can bother him about his horrid creation.
And why blog about such a tiny and tedious topic? Well, the Google coverage of this issue could certainly do with being improved...
I just pushed a beta release of PyFeyn up to PyPI, the Python package index, for the first time. While pretty simple, I'm happy about the whole process - releasing bundled code to the whole community should be this easy! Here it is:
http://www.python.org/pypi/pyfeyn/0.3.0b1 There are several things to do before releasing a non-beta 0.3 version, such as providing some constants for colours and thicknesses so that PyX really isn't needed unless you want to be clever. After that release, there's still room for tidying up some aspects of the interface and the ways the packages interact. On a more flippant note, I like the idea of lines being able to have "haloes" - this would involve drawing a thicker version of every object in white (or another halo colour) before drawing the main diagram, so you could have a neat white border around the diagram elements when putting the diagram on a dark background. Like most fun ideas, it's probably completely useless :-) And I suppose we'd better work on some documentation, too...
I just spent a couple of hours wondering why I suddenly couldn't use files belonging to a particular group on my Linux system in work. For example, a directory is group-owned by a group called rivetgun, it's group-writeable and calling groups tells me that I'm a member of that group. But I can't write to it... what's up? The permissions of the group look identical to those of other groups that are behaving as normal. Eventually, I try logging in as another member of the same group... it works fine! Okay, so it's my problem, not the group.
For ages I've wanted to have a network attached storage (NAS) system in the house, so my music collection, photos, reference documents and so-on can be accessed from one place. Last week I took the plunge and bought myself a new LaCie Ethernet Big Disk 1Tb - a whopping terabyte of NAS goodness. In the last five minutes, though, I just turned it off and packed it back in the box to be flogged on eBay. Here's a brief description of the one week voyage from happiness to disillusionment.
First, when it arrived I had much the same warm fuzzy feeling as when my Mac Powerbook arrived a few years back: it's solid, silver, beautifully presented: full marks to LaCie on quality of design and construction. The user manual is a little thin, though, and despite saying "fully Linux compatible" on the side there's no mention of Linux in it. Hmm, warning bells.