Re: Gabor resigns from the editor post

Rasmus Lerdorf wrote: >>> So, as you can see there are a lot of repeated results. A simple >>> solution is to search only for urls with /manual/$LANG/ on it. >> >> Not really. We encouraged users and search egines to get to know and >> fall in love with our short URL scheme. Since search engines utilize URL >> canonicalization and they end up with our short URL forms, restricting >> searches only to the /manual/$LANG pattern would exclude a lot of manual >> pages. We should rely on the fact that manual pages output a >> Content-language HTTP header (which seems to be proper to me, Rasmus had >> some reservations recently about them). > > I did? I don't recall that. But I am checking with the search gurus > here on the best way to identify a page's language to the indexer. I > have a feeling it isn't listening to this Content-Language header. Maybe I did explained it badly. I was refering to this quote from you (from yesterday): | For this to work we really need to specify the language of the | manual pages in a header at the top. Otherwise the search engine | has to guess, or worse, we need to manually filter the results. | We know the language when we generate the page, so it makes sense | to tip off the search engine. Sometimes it is really hard for | it to guess because a page may only have a couple of non-English | words followed by lots of English in the notes. If Content-language does not work, then it might be because the indexers try to be smarter then the HTTP headers. Goba