Seth Wagoner’s Blog – Covering Interesting events in his life and our times.

Following up on my previous post about the Christchurch Digital NZ hackfest, I did find the bug in my search plugin script, it was just one of those minor typos that takes ages to find because the error message you got when you tried to use it was completely uninformative. Sigh. However, since I spent most of my time on this during “Mozilla Service Week” I chalked up a few hours there to add to their total. Kudos to Mozilla for organising that and I’ll be sure to take part in a more serious way should they do it again.

Digital NZ has created a Custom Search Builder, and it seems to me that they could add Search Plugin generation to this reasonably easily. All they need to do is take a copy of my sample and use it to create a template the swaps out the content of the ShortName and Description tag, and also the template attribute in the <url> tag – everything else can stay the same. Then they need to add a javascript install link to the search result pages similar to the one here.

Heres what my sample search plugin file looks like (NB: First bit of code I can recall posting in this blog. I promise not to make it a habit.)


<OpenSearchDescription xmlns="http://a9.com/-/spec/opensearch/1.1/"
xmlns:moz="http://www.mozilla.org/2006/browser/search/">
<ShortName>Digital NZ</ShortName>
<Description>Digital NZ Archive Search</Description>
<InputEncoding>UTF-8</InputEncoding>
<Image width="16" height="16" type="image/x-icon">data:image/x-icon;base64,[data string goes here]</Image>
<Url type="text/html" method="get" template="http://search.digitalnz.org/en/search">
<Param name="search_text" value="{searchTerms}"/>
</Url>
</OpenSearchDescription>


The Webserver has to serve up that XML file with a MIME type of application/opensearchdescription+xml

eg using apache, in a directory where .xml isn’t reserved for anything else, one could put this in a .htaccess file:

AddType application/opensearchdescription+xml .xml

It’s also relatively trivial to add something to the headers of a webpage to enable autodiscovery of one or more search plugins. I’ve done that on my example page. Once you’ve done so it’ll appear in the search plugin manager for the browser, like this:

For more details, and how to do other cool things like enabling search completion, see:

OpenSearch Specification
Mozilla article on creating OpenSearch Plugins for Firefox

Fellow Kiwifoo camper Jo Eaton was in town on the weekend, spreading the good word about their mission to index all New Zealand’s digital content, and make the data available via their new developer API’s, with a travelling “Hackfest”. Fortunately It wasn’t too far for me to walk to take part – it was upstairs at CII, where Interclue is located.

Among the various hacks there was an iPhone app and a Drupal module, and I got most of the way through building a Search Plugin for Firefox and IE, which is a relatively trivial hack in theory but I’d never built one before so it was a useful learning experience.

Unlike fully blown browser extensions such as Interclue or Lazarus, search plugins are just an xml file that when loaded using a special javascript method (only available in certain browsers, such as Firefox 2+ and IE7+) will cause your browser to create another search provider for the search box, which by default in Firefox only has a few general purpose search engines such as Google and Yahoo available, and a few site specific ones such as Wikipedia. But anyone can create a new search plugin for the search on their website, and getting users to install it can mean that they come back to your website more often.

I found a few little niggles, such as that the xml file had to be served up with the right MIME type by the webserver, and that the best way to provide the icon was using a “Data URI” – essentially a way of encoding an image using text. Fortunately Hixie has a kitchen for that.

My attempt is here, and in theory it should install fine in IE7+ and Firefox2+ by clicking the link on this page, but so far, it doesn’t, and I’m not quite sure why. I’ll update this post once I’ve fixed it! [update: fixed]

DigitalNZ has a “roll your own search engine” system set up for their growing collection of Digital Kiwiana, and it should be simple enough to extend that system to build a search plugin for each derived engine, since they will share the same pattern apart from the target URLs. There are also standards for search completion (guessing what you want to search for) and autodiscovery. I’ll make another post in a couple of days once I’ve had a chance to figure it out properly.

[updated because I forgot their mission was only to index the metadata, the digitizing and putting online bit is up to the contributors and partner organizations]

As I mentioned, Lazarus is now a Mozilla recommended add-on. That seemed to be good for about 3,000 downloads a day last month. But on Sunday night I glanced at our stats and noticed we’d had 10,000 downloads. Gosh.

Provisionally, I am inclined to blame this tweet, which eventuated shortly after this post. Cheers Fred. :-)

I expected the download rate to tail off after that. However, it’s actually steadily increasing, thanks to a spate of retweets and blog posts from people like Rick Broida, who is syndicated all over the place. We might clock over 100,000 installs for the week at this rate. That would put Lazarus in the the top 20 Firefox addons, maybe even top 10. Unfortunately I can’t tell right now because the AMO stats are broken – the last 5 days of traffic haven’t been processed for some reason. So everyone’s weekly numbers are currently way off. That may be fixed by the time you read this post, of course.

A few weeks ago I was talking to my advisors about whether we should be focused on Lazarus or Interclue. I argued that Interclue had better long term monetization prospects, and showed them my marvelous J-curves. Nat pointed out that Lazarus had the more obvious value proposition and some incredible user feedback. I threw a few ideas into the ring on how we could extend and monetize Lazarus, and it was mooted that it would be best to get at least one of those ideas into action before we started to actively promote it. However, it appears that active promotion wasn’t exactly required in this case. We appear to have entered a cycle where every time someone complains in a tweet, on a forum, or in their blog that they just lost a pile of text they typed in, someone else chimes in and says “Got Firefox? Get Lazarus.” or words to that effect. Hence we have some pretty impressive word of mouse going on right now.

[UPDATE: It appears there was a bit of a stats glitch! We were never getting more than ~5k installs / day for Lazarus. However, ~5k/day is still very impressive and a considerable boost over what we were getting before that]

She’s everybody’s sister… she’s symbolic of our failure… she’s the one in fifteen million who can help us to be free.

“Watching TV”, from Roger Water’s amazing album “Amused to Death“, contains surely one of the more poignant statements of the value of mass media in evolving peaceful democracy that you’re ever going to hear. Given the events of the day, I suggest giving it a listen. (And if you haven’t already, you might want to buy the album, and listen to it on a good sound system, it really is awesome. )

Roger’s Yellow Rose was a student of philosophy. So was Neda Agha-Soltan. So was I.

And I grieve for my sisters.

I’ve made a few updates to my Technoprogressive page. It’s always been little more than a half-baked collection of thoughts, I should really tidy it up properly one of these days. Still, a few interesting nuggets in there if you’re into that sort of thing.