Anthony Bailey

Alice and Kev is a “real life” Sims 3 story,...

Fri, 19 Jun 2009 21:51:39 +0100

Alice and Kev is a “real life” Sims 3 story, told with images from and textual descriptions of unfolding events in that game/world.

The author Robin Burkinshaw set up a homeless father and daughter, and watches the story that has emerged. He makes wonderfully artful choices in his character setting, occasional intervention and narrative interpretation; but the game itself comes across as an equal creative partner.

Fascinating concept. And the tale is really compelling. This is the only soap in my feed reader.

(I have little interest in an ontology debate, but, tag: machinima)

Progression of feelings about meta-programming. Scary and...

Sat, 13 Jun 2009 00:09:23 +0100

Progression of feelings about meta-programming.

Scary and Bad: People are wary of meta-programming and don’t use it much
Scary and Good: people begin to see the value of meta-programming but are still uncomfortable with using it.
Easy and Good: as people get comfortable they begin to use it too much, which can complicate the code-base.
Easy and Bad: people are wary of meta-programming and realize that it’s very useful in small doses.

(From a longer article by Martin Fowler examining the use of Ruby at Thoughtworks.)

Tiny activity

Mon, 25 May 2009 21:33:00 +0100

I recently did a little rewiring of the technologies behind the on-line representation of me. Almost certainly unwarranted given how little watched and how inactive I am, but it means this post gets an apt title, and the approaches were interesting for their own sake.

Firstly, a little vanity. I’m pretty happy using anthonybailey.net as my home on the Net, but it can be incoveniently long sometimes. So I registered antb.me/ as a tiny URL vanity domain, and wired up my blog framing machinery so that I can shorten e.g. http://anthonybailey.net/tumblelog/2009/03/30/a-happy-accident-memetics-and-pairing to antb.me/t214 if I want to.

Secondly, some aggregation. I display a headlines summary as part of my homepage. This used to be generated from a special-purpose merge of my blog and tumblelog feeds, but I am a little active on sites that I don’t own as well, and activity streams are all the rage. So, now it’s powered by a FriendFeed blend of these blogs together with comments tracked by the excellent Disqus and BackType services. (This is still imperfect. Ideally, FriendFeed wouldn’t abbreviate my blog content and then the feed itself would be a more useful read. Actually, ideally trackback would have caught on enough that I could comment on other blogs by posting on my own, but that wishful distributive thinking just hasn’t worked out.)

Space Elevator Pitch

Thu, 21 May 2009 14:21:19 +0100


  space elevator pitch
  
    When a "great new business idea" takes
    so long to describe you know it can't work.

Dollhouse is well worth watching. The first season of Joss...

Tue, 19 May 2009 00:43:05 +0100

Dollhouse is well worth watching.

The first season of Joss Whedon’s latest TV show ended in the US ten days ago, got renewed at the end of last week. I plan to write a proper review at some point, but wanted to give two thumbs up in the meantime since it starts on Sci-Fi here in the UK on Tuesday evening.

"Uncle Bob is trying to create a world where nobody ever got fired for writing a unit test."

Sat, 09 May 2009 15:39:46 +0100

““Uncle Bob is trying to create a world where nobody ever got fired for writing a unit test.””

- Giles Bowkett worries about overprofessionalism in software development. I think the warning is way premature, but I like the phrasing.

Buzz Out Loud discuss a partial brain simulation.

Fri, 24 Apr 2009 22:28:25 +0100

Tom Merritt: [The researchers] can trace back every activity of every molecule in every cell, every connection and see how the memory was formed... in ten to twenty years they may be able to create an entire brain
Natali DelConte: Essentially then you upload experiences in the same way you do in the matrix, you can just plug in and... then I'm going to know ju-jitsu and you should all just watch out.
Tom Merritt: The thing is, you won't be able to know whether you're in a real brain or not... they could just make infinite copies of you... and how do you know you're you any more?
Natali DelConte: Why do I care? *I know ju-jitsu.*
(from Buzz Out Load podcast #958.)

Continuous deployment is a provocative phrase. The recent blog...

Thu, 23 Apr 2009 09:45:15 +0100

Continuous deployment is a provocative phrase. The recent blog post by Timothy Fitz may have produced as much heat as light and the inevitable “yeah, but we can’t because…” but the concept provokes good conversation in this Technometria interview (54min) where he and Phil Windley discuss the underlying driver (basically lean product development involving the end-user earlier) and nearby thoughts such as YAGNI in end-to-end test design, lessening the pain of load-testing in production, managing intermittent test fails, etc.

Is the Supremacy of OOP Over?

Tue, 21 Apr 2009 23:31:34 +0100

Is the Supremacy of OOP Over? : Thoughtful article advocating less ceremony and fewer objects over at (irony!) Object Mentor.

"There are no silver bulletproofs"

Tue, 31 Mar 2009 21:00:13 +0100

“There are no silver bulletproofs”

- Rory McCune narrowly avoided this pun in his Scotland on Rails talk when explaining that a web app framework may help in avoiding security issues, but it can’t solve every problem for you.

Coding, Super, Way, To, Object, Kernel

Tue, 31 Mar 2009 00:38:57 +0100

Some of the most enjoyable parts of Scotland on Rails concerned the Ruby object model. Dave Thomas presented a keynote that made the basic model crystal clear. (If the videos aren’t available, I expect this screencast covers the same content.) Earlier I’d attended the charity tutorial which explored more of the details and consequences, with plenty of fun examples.

Here’s some poetically obfuscated code I wrote this evening that plays with all of the ideas.

And here’s the output in Ruby 1.8.6. (Ruby 1.9 probably mentions BasicObject too.)

Can you figure out how it works?

A happy accident - memetics and pairing

Mon, 30 Mar 2009 00:53:00 +0100

In his Scotland on Rails 2009 keynote, Michael Feathers was describing how coding style can be passed down a chain of developers: “I learnt that approach from a team mate who worked with a guy who worked at MIT.” He noted that such “memes” often die in industry: “yeah, Chris over there in the corner loves that functional style - the rest of us leave her to it.” Shortly after he happened to observe how surprisingly wide the adoption of XP technical practices has spread.

This coincidence in talking points was a happy accident for me that provoked an “aha!” moment relating memetics and agile practices.

The “meme” concept is itself widespread, albeit quite diluted from its origins. Back in the day, I had an interest in the more formal notion of memetics; here’s Wikipedia’s definition:

Memetics is an approach to evolutionary models of cultural information transfer based on the concept of the meme. Starting from a metaphor used in the writings of Richard Dawkins, it has since turned into a new area of study, one that looks at the self-replicating units of culture. It has been proposed that just as memes are analogous to genes, memetics is analogous to genetics.

Ideas spread based on how persuasive they are when you hear them, how much you want to pass them on, and how often you have the opportunity to do so. Fortunately, truly valuable ideas tend to do OK under these conditions - when ideas actually work, they do tend to prosper unless they are very hard to communicate. But also some kinds of ideas get an advantage orthogonal to whether they actually work: ideas that unduly bias the host to propogate them, and ideas that give the host more opportunity to persuade. Chain letters are a well-known example of (the phenotype of) the first meme. A belief system (Mormonism, say) that advocates having large families is an example of the second; the more children you have, the more chance you have to pass down your faith.

Now, it strikes me that pairing (and other practices advocating team communication) have a memetic advantage of the second sort. As well as being plain dumb, a “never talk to other developers” meme would clearly have had less opportunity to spread.

Another concept from the field is that of the “meme complex”, analogous to a gene complex: a set of ideas that travel better together that they might independently. For example, the memetic advantage of evangelism (“tell your friends!”) can partner with the one coming from discouraging dissent (“heretic!”)

Kent Beck observed that the principles of XP are mutually reinforcing. And I believe XP forms a neat meme complex that benefits by riding on pairing’s coat-tails. TDD is a particular beneficiary: a pairing partner is so often the source of test-infection.

I want to be clear that I’m a firm believer in the genuine worth of the practices in question. When I pair, I am happier and I learn much more. My considered experience is that the resulting code compares very favorably with what I produce when working alone. Similarly I’m sure that the teams I’ve worked on are significantly more than the sum of their parts. But I do want to make explicit the observation that there may have been a happy accident: worthy XP practices have enjoyed a memetic advantage that has accelerated their adoption.

[tag: software_development]

I’m enjoying the Scotland on Rails conference. But mostly...

Fri, 27 Mar 2009 23:58:00 +0000

I’m enjoying the Scotland on Rails conference. But mostly this post is because I like this year’s logo - though I don’t know if those from out of town get the reference, or whether to expect any talks on Reverse Polish Notation.

http://academicearth.org

Wed, 25 Mar 2009 23:40:15 +0000

http://academicearth.org:

Over Christmas I really enjoyed watching the videos of MIT’s Structure and Interpretation of Computer Programs lectures. So I was happy to notice an aggregator for this kind of resource: Academic Earth.

I can see my house from here

Thu, 19 Mar 2009 09:17:29 +0000

Fellow locals of Edinburgh, the StreetView images of our fine city are now active in Google Maps.

(I like how the person icon goes all Superman if you drag it around.)

"About twenty years ago people noticed computers and TV were on a collision course and started to..."

Tue, 10 Mar 2009 21:40:00 +0000

“About twenty years ago people noticed computers and TV were on a collision course and started to speculate about what they’d produce when they converged. We now know the answer: computers.”

- Paul Graham on Why TV Lost.

Time It's Time

Tue, 10 Mar 2009 21:09:48 +0000

Around a year ago, a hobby Rails application of mine started responding very slowly, taking about ten seconds longer than the far fewer it used to.

I correctly diagnosed the primary reason: the tiny 128MB virt on which all my stuff ran was getting stressed. I had just implemented some new neat features on the now slow service, one of which introduced some fork-heavy parallelism. So when it came to memory, the virt was now chewing up alloc more than it could byte… sorry.

I made changes that I thought should have helped, but performance never got back to how it had been.

But this was not surprising; other conditions had changed too. I had started hosting a third service of my own on the overloaded virt. Other tenants were definitely using more of its resources. The installed MySQL version had changed, and the DB itself was slowly growing, and my mongrels were configured differently - I had maybe crossed some swapping threshold.

And, I didn’t care much that this service was troubled. My main use case for it had moved into the office at work, where it was running great on rather less conservative hardware. So I let it this instance crawl and whinge and splutter and fail.

Just recently, I switched my other apps to a new host. I expected the free memory to bring the slow one back to life. Nope. Perhaps I’d never really sorted the forking issues, so I simplified that away. Nope. Well, the service did proxy for others that it was supposed to cache asynchronously in the background, maybe I’d messed up and it was still calling them before returning, so let’s just reap the cache instead. Nope.

After removing all the funky stuff I had assumed was broken, the truth was revealed. Simply rendering the view was taking more than ten seconds.

One of the minor neat features I’d introduced way back then involved doing about a thousand simple operations on dates and times, mostly comparing and adding. ActiveSupport makes this stuff very readable. And apparently, sometimes shockingly slow! I tweaked algorithms to avoid most of the computations and I got my ten seconds back.

I would never have guessed that my very simple processing was CPU bound. Live and learn.

Boy is my face red

Fri, 06 Mar 2009 07:08:50 +0000

It’s embarassing when having just filled a hot water bottle you manage to squirt its scalding contents up into your face. (Ow.)

Feed Morsel: Making the Ruby library RSS parser accept the "content" tag

Wed, 04 Mar 2009 00:43:57 +0000

(Here’s a morsel of info I encountered when recently tinkering with feedsum.)

Ruby’s standard library includes an RSS parser for various flavors of feed. But it implements the RSS specs quite tightly, which means it won’t accept the content / content:encoded tags that have become de facto standards in some blogging quarters for holding the entire body of a post. (Others put the whole post in description, but I can see why a shorter summary fits more naturally there.)

The parser comes with clean but poorly documented hooks for defining extensions. Distilling this helpful post, I present the minimal code snippet to do what we need:

class RSS::Rss::Channel::Item
 install_text_element "content:encoded", "", '?', "content_encoded", :text, "content:encoded"
 install_text_element "content", "", '?', "content", :text, "content"
end

RSS::BaseListener.install_get_text_element "", "content_encoded", "content_encoded="
RSS::BaseListener.install_get_text_element "", "content", "content="

If, like me, you’re using FeedNormalizer to wrap your parsing, you might put those lines at the top of your parsers/rss.rb, and update its item_mapping with :content =>[:content, :content_encoded]. Then you can call something like:

FeedNormalizer::FeedNormalizer.parse(open("http://example.com/rss.xml")).entries[2].content

Fliggo offer video-based Ning-like social network hosting

Thu, 26 Feb 2009 21:46:11 +0000

Fliggo offer video-based Ning-like social network hosting:

Individual machinima studios have a large number of options for hosting a video channel. But a broader machinima community wanting to form a stand-alone niche social network are less well-served.

You could write one from scratch, like Machiniplex. (If you wanted to do that today, you might look at using Panda.) But that involves some technical nous.

There are a few platforms knocking around for more easily building niche social networks - Ning is a well-known one. TechCrunch today featured such a platform that specializes in social networks based around videos: Fliggo. Still in development, but out of beta. Looks worth considering if you wanted to get something up and running with minimal effort.

[tag: machinima]