|Problems? Is your data what you think it is?|
RE: Considering Super Searchby brainpan (Monk)
|on Oct 31, 2000 at 05:06 UTC||Need Help??|
Excellent suggestion footpad; I've been wondering if I'm the only one who can actually
end up worse off after a search than beforehand. While we're listing helpful features for
Super Search here are a few I'd like to see:
1. Excerpts from the returned nodes (a la Google).
2. AND, OR and NOT operators. The existing search facilities seem to only do
OR style searches, meaning if I type in five search terms which only
appear together in one or two nodes, but separately in 200+, the existing
search will give me 200+ results. Sometimes this is desirable, but not
usually. My memory is very detailed, but just as selective. I'll remember
three lines of code verbatim, yet an OR style search actually makes
adding more search terms widen the scope of the returned pages, not
narrow it. This is not good.
3. Search only (non)code tags.
This would mainly be a performance enhancing feature. Depending on how
the search engine works this might slow things down though. :\
4. Restrict search to nodes referenced from the thread FOO, descending BAZ threads.
Maybe I'm the only one who does this, but frequently I'll recall where I
started, a few words from the page I'm looking for, and that I
traversed +- 4 threads from FOO to get to the page I'm looking for. This
may be hard to implement (or maybe it'd be three lines of perl; I'm not
good enough to know yet), and I imagine it'd be rather expensive in terms of
computational power (or at least memory).
Feedback people - am I the only one who'd use a feature like this? Would
implementing some or all of the features mentioned in this thread be
enough to render this feature superfluous?