Hi all,
just a quick status update: The dump is currently running at 2req/s and ignores all pages which have is_redirect set; also I changed the storage method: the new files are appended to /mnt/user-store/dewiki_static/articles.tar, as I noticed I was filling up the inodes of the file system; doing the storage inside a tarball will prevent this and I don't have to waste time downloading the tons of files to my PC, only one huge tarball when its done. I also managed to get a totally stripped down version of the Vector skin file loading an article via JSON (I won't release it now though, it's a damn hack - nothing except loading works, as I have removed every JS file... should be pretty till Sunday). Current dump position is at 92927, stripping out the redirects 53171 articles have really been downloaded, resulting in 770MB of uncompressed tar (I expect gzip or bz2 compression to save lots of space though). For the redirects: how do I get the redirect target page (maybe even the #section)?
Marco
PS: Are there any *fundamental* differences between the Vector skin files of different languages except the localisation? Could this maybe be converted to Javascript, maybe $("#footer-info-lastmod").html("page was last changed at foobar")?
wikitech-l@lists.wikimedia.org