j2k wrote:
MediaWiki 1.9.3 (from tags), however, I *was* able to get the error to occur when restricting read access in LocalSettings.php:
$wgGroupPermissions['*']['read'] = false;
Thanks, it worked. Till backup I commented out this line. But the dump still not too good on windows...
Now the dump is okay in the \static directory, but with unicode filenames... Thus none of the links works, because the filenames are corrupt. Is there a way to convert page names to windows compatible ( :) format?
Windows XP has no problem with unicode filenames, it's a UTF-16 native platform. The problem is that PHP 5 doesn't support the UTF-16 interface, it just calls the compatibility functions, which interpret UTF-8 input as CP1252, resulting in garbage. There's a proposal to fix the problem in PHP 6.
It's no problem for Wikipedia since we generate our dumps on Linux, and transfer them to Windows with a tool that uses the UTF-16 interface (namely 7-Zip). You could do the same, or you could hack dumpHTML.php to generate encoded filenames. The problem with encoded filenames is that they are unreadable, so you can't find the article you want by browsing the files on disk.
-- Tim Starling