We notice sites that host automatic updates and consider improvements for each.
We generate sensor reports from flat files maintained within the SensorServer 
system with the mksensor script.
This foreign server is hosted at c2.com and served through apache including a snapshot of the Sep 25, 2011 client.js code which can almost launch today. site
The cited mksensor cron script builds a dozen pages every five minutes but dates only the index with the current date.
update: revised mksensor script to use its own modification data as index's creation date but with last refresh time noted in the page text.
Sites hosted by this farm with activity in the last 10 days.
Sites hosted by this farm with activity in the last 10 days.
This page hosted by a well maintained node wiki is rebuilt hourly by the cron script called farm-scripts. github
The ruby script uses File.mtime to date the journal action for each mechanically generated Reference plugin item, one each for every included site. Thus it inappropriately sees remote updates to any JSON plugin in the farm.
The local clone of the repo has one uncommitted change: clearing the status/sitemap.json file with each update.
Here we report sites we find with recently edited pages.
Here we report all sites found, organized by domain name, excluding sites with less than ten pages.
These pages are part of the foreign server that hosts our slow but still useful search. Each is run from the scrape cron script four times a day. github
update: revised sitemap generation to use script file's modification time rather than run time. github
.
What are the best practices to Feed the Federation? I don't know yet but would like to find out.