[textile]
"blo.gs":http://blo.gs is a good service based on a good concept, and I have enjoyed using it during the past few months. The goods:
* Updates when the individual site send a ping to blo.gs. This must be the best way to know when a site is updated.
* Autodiscovery of RSS and Atom feeds.
* "XFN":http://gmpg.org/xfn/ support
The bads:
* I don't know how to write PHP to grab the xml file and display the blogroll on my site, so I use the "javascript implementation":http://blo.gs/favs.php instead.
* It's huge. And thus could not be access sometimes. And the javascript time-out is too long. On occasions I have to wait a good 30 second and even then it might decides to not load at all.
* Search engines and "Technorati":http://www.technorati.com cannot crawl javascript-generated content.
As a test run, I have taken the blo.gs link out and instead use the weblink monitoring which comes with Drupal's "weblink module.":http://drupal.org/project/weblink What is happening is, basically,
* I added items from my blogroll manually, as if adding just another "weblink":http://ichris.ws/weblink
* Every one hour, a cron job is run, and my site grab the RSS or Atom feed from these monitored sites, and compare them with old copies.
* If there're more than 40 kb difference in a feed, the site is marked as updated.
* 20 most recently updated sites are shown on the side bar. (appears only in blog section)
The system is only in place for a few hours, and I'll have to wait and see if I need to tune any parameters (fetching frequency, change tolerance) to make it more accurate. Also the server load has to be monitored, if I am going to add very quick-update sites like "Gizmodo.":http://gizmodo.com
If it works well, this site will be totally independent of any outside services. (the new tagboard is also run on ichris.ws)
OK. Time for lunch.







Add new comment