Re: Bookmark Queryable

[Joe Shaw <joeshaw novell com>]

Hi,

On Thu, 2006-09-14 at 10:51 -0500, Adam T. Gautier wrote:
> After scanning the code for the MozillaQueryable it seems that the code 
> deals with the mail client but there does not seem to be code for indexing
> the bookmarks.  

The MozillaQueryable code has been dead for quite some time now.  Sadly
CVS doesn't have a way for us to move the (growing) amount of dead code
in Beagle to an attic of sorts, so it can be hard to see what's
functional and what's not.  In general, the best bet is to look at what
is actually compiled in by looking at the Makefile.am files.

> What is the groups feeling on such a queryable?  Some features of such a 
> queryable could be:
>
> * Indexing of all bookmarks
> * Indexing of all immediate links in the HTML of the bookmark's URL
> * Use of bookmark wildcards

Is the idea to go out onto the network and download the pages and index
them?  That makes me a little nervous (see below).  Unfortunately,
without doing that the bookmarks themselves aren't super useful (and
that's why we don't index them today).

> * There is an easy mechanism for removing indexes to keep lucene
> synchronized with FF bookmarks.

It is up to the backend to handle its own removals from its own index.

> * Mechanisms for rendering remote websites is easily attained.  I had
> problems with a basic programatic access to google search results one
> time.  I fixed it by having it completely simulate a mozilla browser which
> was a pain and in java so I hopefully will not have to start from scratch
> with c#.

In general I am rather nervous about indexing remote resources for a
number of reasons:

        * Doing so can be extremely taxing on the network.
        
        * For random resources (like Google search and bookmarks, but
        unlike something that you have an account for, say Gmail or
        Flickr) there are privacy and security concerns because Beagle
        is essentially phoning home.
        
Joe