Hi all,
this is just a little notice that I'm about to commit a revamp of the
geosearch extension soon, later today or tomorrow morning, so if anyone
has a concern this would be the time to speak up.
The rationale is that Google changed the way it indexes kml sitemaps. In
the past, it liked them to address specific placemarks and hence the
geosearch extension produced deep sitemaps down to actual features.
But the way it works now, and that has more chances to get crawled and
assigned high rates, is that the kml pointed out by the sitemap should
be some sort of "metadata kml document", meaning it should contain the
layer's title, abstract, may have some sample placemarks and/or bounding
box, and link back to the actual data.
By the other side, the sitemaps generated by the geosearch module were
broken since GeoServer 2.0.x (not sure about the exact version but way
in the past), in the sense that the urls it generated, besides the
toplevel sitemap.xml, lead to 404 http errors, so there's actually no
geoserver sitemap that could be crawled.
Another reason why the geosearch extension was not working is that the
sitemap it published lived under /rest/sitemap.xml. That is, it is not
possible to access it anonymously, as /rest/ requires authentication,
which the google bot does not, of course. So the new sitemap is gonna
live under /geosearch/sitemap.xml (thanks justin for the solution).
So that's more or less it. I'm planning to commit to trunk, have a demo
instance be crawled, and when having a confirmation that google likes it
backport to 2.1.x.
Any comments are welcome.
Cheers,
Gabriel
--
Gabriel Roldan
groldan@anonymised.com
Expert service straight from the developers