Gallery of Station Websites: Progress
After two weeks of use, the station gallery has had a fair amount of traffic and feedback (bolstered by a write-up in Current and a couple station resource sites. While I've been revising the site all along, I rolled out some more substantial changes this weekend, which should improve the functionality and performance of the gallery.
First, and most obvious, I did a little bit of styling work to take it from barely functional to ugly-but-useable. If anyone has design ideas for the site, please leave a comment. Other web galleries usually severely crop their images, but I'm attracted to the full page screenshots, but that comes with a whole range of other design challenges.
As part of the design work, I've moved the "user-generated" content forms to the gallery view (and eliminated the station-view entirely), which I hope will encourage some conversation. I toyed with the idea of integrating with services like twitter and delicious to provide comments and tagging, respectively, but I couldn't come up with a good way to distinguish between conversation and criticism.
I've been working on a couple ideas for a proper home page to the gallery to help give the site some context besides these blog posts. I'll try to make some more progress on this as long as people believe this could be a useful resource for the public media system.
Shortly after the first prototype, I started migrating data into Solr (using the ruby sunspot library) and earlier this week I added full-text search for page content (which may or may not match the screenshot content). I'm still playing with approaches to crawling station websites to extract different types of pages (schedules, contact, news stories, features, etc) with Anemone, hopefully I will have something interesting in the next couple weeks.
I'm still working out ways to automate (and schedule re-occuring) screenshot updates, which is complicated by the lack of decent cross-platform tools. On Mac OS X, I've been using webkit2png, which has been great, but this server is running Debian Linux and the only comparable utility I've found is wkhtmltopdf, which requires a patched version to QT. Messages queues or workflow engines seem like overkill, so in the meantime I'll do manual updates occasionally.
As always, the source code to the site is available at http://github.com/cbeer/publicmediatech-stations for anyone interested in hacking on the gallery or just seeing how it was put together.