From: Suraj Kurapati <snk@...>
Date: 2007-03-25T18:01:49+09:00
Subject: Re: Rassmalog 3.1.0 - javascript search

Suraj Kurapati wrote:
> Josef 'Jupp' Schugt wrote:
>> * Robert Wagner, 09.03.2007 20:47:
>>> regarding the
>>> search problem: consider integrating a javascript based solution.
>> 
>> the integration than should be done in a clever 
>> way. My suggestion is as follows:
>> 
>> 1. Keep the current search
>> 2. Add a search using javascript
>> 3. Make use of <noscript> in this manner:
>>    a) If Javascript is enabled display
>>       "Search (Javascript enhanced)"
>>    b) If Javascript is disabled display
>>       "Search (no Javascript, deprecate)"
>> 4. Add a small paragraph about why to enable Javascript to the latter
>>    search.
>> 5. On the same page add a piece of Javascript that replaces the page
>>    by the Javascript version of the search.

I evaulated Tipue and several other JavaScript search engines, and Tipue 
seems to be the best choice because it does not require prior 
preprocessing (indexing) of the data to be searched.

However, when I integrated Tipue, it did not seem to offer a huge 
advantage over the browser's internal search mechanism. For instance, it 
doesn't implement any query relaxation -- I'm thinking of edit-distance 
and Soundex algorithms -- so it's hard to find something unless you know 
exactly how it is spelled beforehand.

With the browser's search mechanism, you get to see a much larger 
context of a match whilst searching. In contrast, search engines strip 
away much of the context, which is useful in situations like this -- 
where you're manually scanning for something that resembles what you're 
actually looking for.

> I will also remind (although it may be considered offensive to be 
> reminded of such basic operations -- like someone telling you how 
> shoelaces are tied) the reader gently about using their browser's search 
> mechanism, be it <control><f> or <forward-slash>. 

I decided to omit such a reminder, simply because it might seem 
offensive.

> I will also state 
> clearly that the "search" page is really just an exhaustive listing of 
> the content of all entries available on the blog.

I have done this with great reluctance, because it caused the addition 
of yet another template file (config/index.erb), in the new 3.2.0 
release.

-- 
Posted via http://www.ruby-forum.com/.