Last modified: 2010-05-15 15:59:47 UTC
Created attachment 4105 [details]
patch to Title::getFullURL() to prevent building URL with two slashes in a row
The maintenance/generateSitemaps.php script can build URLs with two slashes after the domain.
If using really short URLs (i.e. http://domain.tld/Article), the double slash means a browser is brought to an article with a title like "/Article", effectively preventing search engines from properly indexing content.
The double slash doesn't seem to cause a problem with normal URLs (i.e. http://domain.tld//index.php?title=Article).
Have not tested other short URLs (i.e. http://domain.tld//wiki/Article or http://domain.tld//index.php/Article).
It looks like you have a bogus $wgServer setting.
$wgServer needs to be set to the URL protocol, host, and port (if necessary), with *no* path component, as paths are appended to it.
eg 'http://example.com' or 'https://custom-secure.example.com:4443'.
I feel dumb; it's because I've been calling the script with --server=http://domain/ (note slash at the end)...
So how about a note in the help string about leaving off the slash, or trimming it off of $options['server'] before $wgServer is set to it?