On Wed, Dec 24, 2008 at 04:23:39PM -0800, Daniel Burrows wrote: > Say, oh, for instance, that I want to find http libraries for c++, > so I search the apt Xapian database for "c++ http". It looks like > every single package with "Homepage: http://example.org"; is indexed > under "http", so I get back a whole pile of irrelevant stuff. > Presumably it would be fairly straightforward to ignore URLs when > indexing packages, or at least to ignore "http" in "http://";.
Hello, sorry it took me a while to get my act together on this. I tried now an "axi-cache search C++ http" and the results don't seem too bad. The Homepage: field is not scanned, so the problems shows only when the Description: field contains URLs. However, I reckon urls in descriptions should be moved to Homepage fields. Is this a bug to fix in the indexing, or should it just be a bug to fix in the descriptions themselves? Ciao, Enrico -- GPG key: 4096R/E7AD5568 2009-05-08 Enrico Zini <[email protected]>
signature.asc
Description: Digital signature
