On Thursday 28 March 2002 15:14, you wrote: > Server http://..../activepage/arg1/arg2/ > ^ note this slash... > > to indexer.conf, then mnogo do not want to index the links like above > (i.e., there is now entries in the apache access.log for these links). > > If I put a > Server http://..../activepage/arg1/arg2 > > row, then it wants to index the (incorrect) > ".../activepage/arg1/f608fb25dfde33ccbb5fe5ee7a5e69e9/", instead of > ".../activepage/arg1/arg2/f608fb25dfde33ccbb5fe5ee7a5e69e9/". > > I could write the page to create absolute urls, but I do not understand > mnogo's this behavior. The problem is, that I tried to simulate a > similar situation with static pages, and there was no problem like this.
Looks like mnoGoSearch has a very broken HTTP implementation. What is happening, most likely: http://..../arg1/arg2/, mnoGoSearch thinks 'must be a directory[0]', and fetches http://..../arg1/arg2/index.html, oops, next http://..../arg1/arg2 will yield a 302 redirect *which it appears to ignore* after which it will see a link from that page to f608fb25dfde33ccbb5fe5ee7a5e69e9 and interprets that (correctly if it had reached this point without the redirect) as a link relative to arg1. Any change of using a working external page fetcher with mnogo? Emile [0] Which is terminally stupid. The web has no directories, URLs are, after the slash following the hostname, just indexes into the documents to be delivered. The calling application should make no assumption about the 'meaning' of an URI, because there is none. Tacking on 'logical' stuff to an url... don't get me started. --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
