With the release of the Creative Commons Sherlock channel, it seems prudent to document some of the development process. The Sherlock Developer Documentation implies that developing a channel is easy as pie. As a developer unfamiliar with the tools, I found the process any but straightforward. In retrospect certain things make sense now that I have a working model, but I’m still no expert.
First, the sourcecode for the CC channel is available in CVS at the CC Tools SourceForge.net project. The module is named sherlock. This may serve as a good jumping off point for someone looking for a working example.
Finally, things I wish I had known when I started writing my channel:
When saving your NIB in Interface Builder, make sure you save it in either “Pre-10.2 Format” or “Both Formats (Pre/Post-10.2)”. If you save it in “10.2 and Later Format” Sherlock will refuse to load your channel.
XQuery has great XML dissection utilities, but has a really different paradigm than other programming languages. It can’t (so far as I can tell) read directly from the Datastore (which stores the values of the Interface widgets), so everything you need must be passed in as an input. Anything you need to set needs to be returned in the dictionary.
If you want to add more than three items to an NSPopUpButton you need to use the Menu items palette in Interface Builder. This may be obvious to others, but I spent quite a bit of time struggling with figuring out how to initialize the list of options correctly (and get the value back from within XQuery).
Finally, everything is XQuery is an assignment. Well, not everything, but more than you’d expect, so you may as well start thinking that way. For example, an if...then...else (and yes, the else is mandatory) block is actually more like a ternary construct than a decision block. That is, instead of the following:
if ($foo > 0) then let $bar := 5 else let $bar := 10
You do this:
let $bar := if ($foo > 0) then 5 else 10
Additionally, the following resources may or may not be useful to you if you’re contemplating writing a channel.
In conclusion, writing a Sherlock Channel was a lot more painful that it felt like it should have been. But I’ll be back; there’s work to be done in the CC Search Channel, and I’ll continue to post my experiences here.