Posted by mattc at Feb 28, 09 10:12 AM
... Comments (0)
I'm very busy at work. Things to launch, things to convice people about, blurgh...
Consequently most of my RSS reading that used to happen in quite moments during the office has had to be done during my 90 minute commute. I can bookmark to delicious from iPhone's Safari, but 3G is patchy, slow and Safari isn't a nice way to consume the Intenet when you are mobile unless you long for the days of 28kbp/s dial-up.
Newsgator, the RSS reader, solves this problem - fetching feeds, and sync'ing them across all my different devices, letting me read them without having to surf - but creates a new problem in that now most of my RSS reading is done this way my delicious feed is withering. Newsgator lets you bookmark things (they call it 'clippings') but the API to this is a bit rubbish.
I like delicious, it's plugged in to various other pieces of me - my social graph, this website etc. I don't think anyone really uses delicious as a bookmark tool (search wins over manual recall every time), it's just a link consumption log for people with bad memories.
To cut a long story short, I've had to grab my Newsgator feed, diff it against the previous days to extract the new posts, then cross-post these to delicious. Too much hassle, but at least my delicious feed is now representing my daily consumption again.
I like this. Lets you describe your UI tests in plain english as behaviours your program your exhibit. Possibly the way the tests are written are a bit contrived if it's just for engineers, but good for normal humans.
Pretty amazing visualisation. It could do with a bit of curation to help explain key points/people ('who was Mike?', 'when was r0.1', 'why the big explosion of activity on Oct 20th?' etc.), a bit like the Python example on the code_swarm site.
Perhaps it could read & display some conventions from the log messages or release notes, but curation is quite a personal thing and only really doable to those with a intimate knowledge of the project, so here it goes...
Looking through our logs, mid-to-end of October marked the end of the technical prototype, where Al left and the project started in earnest with designers, flash developers, project managers all starting to make commits, which explains the flurry of activity.
Matt H, the main developer is a sort of persistant floating presence, he's joined by Darren in mid-November who replaced Al. Mike & Tom arrive for a few weeks then gradually fade away as they were contracted in to complete specialist parts of the build.
Pierre, the senior designer, fades out as the build goes on, indicating (in our department, at least) design is very much an up-front activity with lesser input the longer the project exists. As we start v2 of the project around April I'd expect the design activity to increase a lot more.
There's a nice long period of inactivity for xmas :)
The very tall lines in the histogram along the foot of the video are external dependencies (flex, phpdoc etc.) rather than huge commits by a single developer, otherwise they might indicate a developer has been saving up code on their working copy and committing it all at once, which of course would be naughty of them.
Peter B stays too long in the visualisation for my liking - he only committed once, and that was a mistake (we share our repository with the rest of the BBC) - so we should probably filter him out of our logs :)
I expect the project to settle down in to fortnightly bursts of activity to match our release pattern as we work toward a live release.
uh oh, this is how CMS's happen. It's starts off all nimble and helpful, but by version 2 it's grown to Enterprise level and you've hired a sale team.
I used to like writing these little systems. My favourite one (called sub-area generator, aka. SAG) powered large chunks of the bbc.co.uk/cult site and it was pretty much a csv2html code generator with a bit of extra cleverness. People warm to technology like this, like a hammer after you've been smashing nails in to wood with your fist, it lets them extend their usefulness to the organisation, most technology is dictatorial.
Proposes that the usability of (electronic) maps should mirror the way we might sketch out directions by hand to a friend. Generalised, not to scale, scribbled notes, features on the landscape. Boxes and arrows. Google/MS/SatNav et al. fail to understand this. I'd rather my SatNav said, 'get on the motorway, take the junction after the red McDonalds sign, then you'll see a big glass building, head towards that.", then perhaps interject if I get lost.
Computers are too literal, that's not how humans interact with each other, there's advantages to our human imprecision. Some kind of OpenStreetMap RouteWiki could carve out a niche here.
