Xapian discuss - Oct 2011 - reindexing

Hello

I just jumped into xapian, so my question could be quite basic, but I
haven't found anything really useful in the list.

I'm using xapian to make a fulltext search on our local servers,
including some pdf's and doc files, but also our own filetypes. Anyway,
I managed to write an indexer in php and it works fine.
My problem is however, that I have to reindex everything when just one
document is changed. Also deleted files aren't kicked out of the index.
How can I just index the changes? I read something about these external
unique id's, but I don't quite get it, because I would need this id
before doing anything on the database, which means I would have to
search the database to check if a file was updated? Or do I have to
store the indexed files'id's in an external db to check for changes?
This seems quite strange..

Any help would be appreciated.

Alain

On 24 October 2011 08:14, Alain Denzler <alaindenzler at gmail.com>
wrote:
> I'm using xapian to make a fulltext search on our local servers,
> including some pdf's and doc files, but also our own filetypes. Anyway,
> I managed to write an indexer in php and it works fine.
> My problem is however, that I have to reindex everything when just one
> document is changed. Also deleted files aren't kicked out of the index.
> How can I just index the changes? I read something about these external
> unique id's, but I don't quite get it, because I would need this id
> before doing anything on the database, which means I would have to
> search the database to check if a file was updated? Or do I have to
> store the indexed files'id's in an external db to check for
changes?
> This seems quite strange..
I think the FAQ on unique IDs should help here.

http://trac.xapian.org/wiki/FAQ/UniqueIds

Handling deletions is a bit harder; you'll need to run through the
database, checking if each document still exists.

If that doesn't quite explain it, feel free to ask again.  Also, if
you manage to work it out, but there's a bit in the documentation that
could be clearer, we'd love input on that - we're trying hard to
improve the documentation, and have a sprint coming up this weekend to
improve the documentation.

-- 
Richard