Re: [Tracker] [TRACKER] Info

[jamie <jamiemcc blueyonder co uk>]

On Wed, 2007-03-14 at 17:27 +0100, Andrea Caccin wrote:
> 2007/3/14, jamie <jamiemcc blueyonder co uk>: 
>         On Mon, 2007-03-12 at 11:51 +0100, Andrea Caccin wrote:
>         > Hi at all,
>         > I'm Andrea Caccin, a student of Computer Science in
>         University of 
>         > Padua, Italy.I have some question for you:
>         >
>         > 1. I'd like to know where can I find the source code of
>         ranking
>         > (frequency, weight) algorithm.
>         
>         we will be moving towards IDF. at the moment its just weight *
>         freq 
>         
>         
>         > 2. what's the maximum byte dimension of collection indexing
>         by
>         > Tracker?
>         
>         ?
>         
>         
>         > 3. what's the maximum number of documents indexing by
>         Tracker?
>         
>         2 billion
>         
>         > 
>         > Thanks in advance
>         > Andrea Caccin
>         > _______________________________________________
>         > tracker-list mailing list
>         > tracker-list gnome org
>         > http://mail.gnome.org/mailman/listinfo/tracker-list 
>         
>         
>
> Thanks a lot for the answers, but in the second I would to know if
> there's a bound (in MB or GB) of the dimension of the documents
> indexed.

there is no size of document limits as such

There is an overall size limitation in sqlite which can be found here:
http://www.sqlite.org/faq.html#q10

we compress the fulltext with zlib in sqlite so bear in mind size limits
are the compressed size - the uncompressed can be up to 10x bigger than
that (up to 10:1 compression)



> I also would to know, if it's possible, the specific file that runs
> the ranking.


the rank as stored in the index can be found in tracker_parse_text : 
g_hash_table_insert (word_table, index_word, GINT_TO_POINTER (count +
weight));       

in
file :http://svn.gnome.org/viewcvs/tracker/trunk/src/trackerd/tracker-parser.c?revision=539&view=markup