I’ve heard about PhotoSynth, Microsoft’s rather cool technology for placing photos within a 3D space, but this Photosynth demo by its architect Blaise Aguera y Arcas is truly impressive – listen to the gasps from the audience as the penny drops on what they are seeing.
I’d love to throw it at some Geograph squares – some of the more touristy squares have oodles of PhotoSynth fodder.
I’ve been using PHPEclipse for PHP development for 9 months or so and finding it a real time saver. Sadly it seems that further development has stalled, possibly because of Zend’s announcement in late 2005 that they too would build their own Eclipse plugin.
That has finally borne some fruit with the recent release of a reasonably functional version of Zend PDT, or PHP Development Tool.
Snappy name.
Still, even though a final release isn’t slated until september, this version isn’t too shabby and appears to have a comparable feature set to PHPEclipse already. I haven’t tried the debugger yet, but editing features seem good, with folding of comment blocks and methods, help with PHPDoc tags, nice class inspector pane and reasonable parsing of php code for problems (it doesn’t seem to spot as many problems as PHPEclipse, but maybe I’ve missed a setting somewhere).
So far, it’s shaping up to be pretty good IDE, particularly if you are already using Eclipse. Remote debugging of PHP, in combination with Subclipse SVN plugin and the Web Tools Platform plugin, gives you a pretty capable development environment.
Methinks I’ll be switching permanently soon!
A number of people have emailed or left comments frustrated by the speed of pastebin. Over the past few months I’ve spent the odd ten minutes here and there tweaking it in the face of ever-increasing load. It’s clear the software just isn’t up to the task anymore, which was particularly evidenced by a recent slashdotting.
Over the next few days I’m rewriting the internals to be much faster, and hopefully release it over the weekend.
Wish me luck!
Went to another meeting of the Hertfordshire Linux User Group on Wednesday night, when Rob Davis gave a talk on his experiences taking the Linux Professional Institute level 1 certification exams.
While I’m not sure that the certificate itself will carry much weight with employers, what is interesting is how studying for the certificate forces you to investigate areas you’ve not previously explored, rounding out your knowledge.
The only cost is the exam itself, which at the test centre closest to me would cost £78 per exam (US$150), and you need to sit two exams for each level of certification. Should be an easy task to persuade an employer to stump up such a small amount for getting more knowledgable employees.
There are some books around to help too. Rob read LPIC I Exam Cram 2 and wrote a review of the book on Slashdot, but (sadly) his (admittedly profusive) use of (too many) parentheses drew a few less then complimentary comments. I noticed Amazon also have an O’Reilly book published in August which I guess would be more to date too – LPI Linux Certification in a Nutshell, 2nd Edition.
Here’s a neat idea – a Google Earth KML file of geolocated wikipedia articles. As you roam the Earth, see links to relevant articles popup. Might be even better as a dynamic link, so that Google Earth simply requests articles for the field of view, then it could stay updated, but still, a nice concept!
Way back in May I wrote an entry about possible problems faced by Geograph’s planned use of “closed” content like Ordnance Survey (OS) maps in combination with our use of Creative Commons (CC) licences.
There are a few areas we need to be sure of before incorporating OS maps into the site:
Thorny questions, and the OS have (quite rightly) spent some considerable time pondering them. Our overriding concern is “protect the archive” – we want to ensure the archive can continue to grow and remain free (as both in speech and in beer) forever.
I think we’re almost there…
Technically, the OS “own” the national grid system, and their right to refuse to allow someone to use it has been upheld in recent case law. Since the project is based around the grid system, to find we are unable to create geolocated works using national grid references would be something of a showstopper. Thankfully, the OS regard using a National Grid Reference (NGR) to indicate a point of interest is a legitimate use over which they can have no influence.
So it seems the OS is happy to allow us to create CC licenced works including an NGR. Even if they weren’t, we could switch to maintaining an archive of works where the geolocation data was a WGS84 latitude and longitude, and then simply derive an NGR from that ourselves. In the event anyone “came after us”, the original archive can remain intact, just our derived works become voided.
Thankfully, it seems we are on safe ground.
I covered this one in my May article – the resulting page as a whole is not a derivative. The page is a collective work, and we’re free to include content on that page using different licences. The individual image remains CC licenced, but the entire page content is not.
Here’s where is gets surprising. After several months of thought, OS have said “yes”.
Is that the sound of jaws dropping?
OK then, it’s “yes….but”.
If someone were to attempt to recreate a map using geolocated metadata from the collective works, that person would be leaving themselves open to legal action by the OS. If our theoretical defendant produced Geograph-derived works in his defence, the OS would argue the geolocation had been derived from their property.
In addition, Geograph itself would not be held liable for the action of this third party.
So, on the face of it, it looks good. The cloud in that silver lining is that building maps from metadata in Geograph images is not recommended. Even something simple like extracting all photos of churches with a 10 figure NGR, and then submitting them as points-of-interest to OpenStreetMap could be (in the words of Egon Spengler) “bad”.
I’m writing this to hopefully spur some more debate on our use of Creative Commons licencing. Our aim is to support the maintenance and growth of a free and useful photographic archive. Ideally forever.
If we can get through the next few decades without the services of a team of lawyers, I think we’ll manage it.
In a recent post about data archival, Ed Parsons linked to an interesting piece about how the 1986 Domesday Project was rescued.
The original project was distributed on laserdisc and accessed with a BBC Master micro, but is now available online at www.domesday1986.com. The interface is a copy of the original BBC interface, and looks archaic to the hypertext-savvy web generation. I believe there are contractual reasons for keeping it that way:
Last year I got in touch with George Auckland at the BBC who was involved with the original production. I enquired whether there was any possibility of incorporating the data into Geograph. While he was excited by the prospect, it turns out they don’t have the rights on the content beyond the original Domesday Project
because of “the rather specific way the project was referred to in the paperwork”.
Looks like we might have to wait until the 22nd century to make something more useful out of that data, but I (and George) live in hope that it will be sooner than that.
Although Geograph is backed up nightly to multiple locations, we’ve a plan to make an “archival” version available via bittorrent, which would be the JPEG images and XML based metadata. We’ll burn copies to whatever is backup-media-du-jour too. Hopefully we’ll ensure it is preserved for future generations!
We’ve tried contacting the British Library a number of times to see if they’d be interesting in holding an archival copy for their Digital Object Management Programme, but they never reply to our emails. You’d think a snapshot of Britain and Ireland comprising 250,000 photographs would be worth hanging onto….
Interesting post on the memcached mailing list about how Digg are using Memcached for session storage. Memcached is intended as a caching layer, but after their session database crashed, they simply went with using Memcached for session storage period.
The new Geograph setup should wind up with around 10GB of memcached space across 5 nodes, and I was hoping to do something similar for session storage. It sounds like the Digg system barely broke a sweat doing this as their memcached nodes were already being hit much harder by their main application.
If it’s good enough for Digg, it’s good enough for Geograph!
I’m rarely “off the grid” but while away camping over the weekend the pastebin database seems to have suffered a pretty catastrophic failure. For the time being I’ve reset it while I investigate…
Thanks to the (many!) folks who alerted me!
Edit: OK, I’ve restored a backup from a few weeks ago, which means most recent posts are gone. Many apologies for the inconvenience caused, still looking for the root cause (all the pastebin tables were just *gone* 
).
The new servers for Geograph arrived this morning, behold the almighty Tower of Power!
The supplier for this lot was World of Computers and they’ve been very helpful in my dealings with them. They were recommended to me and so far they’ve been great. They even deliver the stuff themselves in their own van. I like that. It might cost them a bit more, but they can be sure it’s taken care of in transit and is received by the right person. Nice 
Geograph is a spare time project and I’m on an absolutely monster schedule for the next few weeks to launch a new product, so it could be several weeks before we get these babies racked up. I’ll write more about the architecture of the new setup as I do it…
