Chris proposes a search system for gentoo.org: 1. linked prominently from the front page 2. indexed hourly 3. includes all mailing lists and web documents I'd be willing to help, although I have no experience implementing search software and don't know what's available. (I'm more in favor of a software indexing scheme for the mailing lists and web site rather than a third-party or google approach. ) ---- Dave says: Are there any plans to provide some sort of search for the gentoo mailing lists? I suggest using google http://www.google.com/services/free.html its customizable and free, though no guarantees are made as to what on the site gets crawled. Right now google says they have 30300 pages for gentoo.org and 8310 pages for lists.gentoo.org. NetBSD uses this as their search engine and I find it to be pretty good. Dave
sherman says: I like the google idea. It looks like it works well for netbsd (http://www.netbsd.org/) and should take very little time to implement.
I think the google idea is not entirely accurate. The problem is that google won't completely crawl the entire site unless you pay them for the service. Also, the mailing list archives will not be completely indexed. Then comes into question the matter of updates, which google doesn't do frequently unless you pay them. I'm sure that the volume of duplicate questions on the mailing lists could be lowered with a proper search engine feature. The original suggestion is a good one and extremely *easy* to impliment. Heck, gentoo's distribution contains all the software you need. Just emerge htdig, set up some parameters and bam, you have a search engine. As for the mailing list software, mailman, there are two patches available: http://sourceforge.net/tracker/index.php?func=detail&aid=444879&group_id=103&atid=300103 http://sourceforge.net/tracker/index.php?func=detail&aid=444884&group_id=103&atid=300103 These provide the interface for htdig as well as provide increased accuracy in the results. It even puts the search boxes on each of the mailman webpages. I don't think this is asking much, as it would involve 1 hour (tops) to impliment. Remeber: Google is great for generalized searches, but specific site searchs really ought to be serviced by a dedicated search engine. Just because BSD does it that way doesn't make it right. I can name numerous linux distros which take the approach that I'm suggesting. On a side note, you might want to provide these patches in the mailman package available via gentoo. I'm sure more then one person would find it handy.
Website now has search engine(thanks to Daniel Robbins), closing bug :)