Re: Client-side searching proposal

Rob Hartill (hartill@ooo.lanl.gov)
Mon, 30 Jan 1995 19:20:25 +0100


>
> There is one thing wrong with this proposal: it assumes that the
> entire document can be sent to the client,

The search engine can be smart enough to only add a client-side search
extension to the URL if it makes sense.

> and this will become less
> true over time... there will always be a need to do searching on the
> server side.

Search engines will still run on the server side, but they
will be able to give the client a list of URLs with the extra
search info attached. If the server is asked for only part of
a document the extra search info in the URL may still be of use, but
if it isn't - it doesn't matter.

> Personally, I think the whole # and ? thing needs to be rethought.
> With any kind of complex query, the URL's get very very ugly, and very
> long. Something more like the old TEXTSEARCH method, done on the
> server side might be a whole lot better. The server could send back
> hit occurence data in the headers returned with the retreived text,
> and the client could use such data for highlighting, and also
> scrolling purposes.

Yes, this is possible. The original proposal however would also allow
anyone to specify a cleint side search keyword, which would mean
that I could jump to and highlight any part of your document that
I wanted, without you providing a search engine or adding an anchor
specifically for my needs.

rob.