Current short term and long term goals

• Change property names: Till now, beagle allowed the Filter and Backend authors to use whatever property names they wished. This has created duplicate, redundant and inconsistent property names making it extremely difficult to write a generic frontend for beagle. This is discussed in detail in Properties Hack Week.

• Update Lucene: Currently beagle uses Lucene.Net-1.9.1. Update to Lucene.Net-2.1 which has several useful improvements. In fact, Lucene-2.2 has impressive performance enhancements but it is not yet ported to C#.

• Use stored procedure for the Sqlite operations in beagle. This should speedup textcache and fileattribute-store performance.

• Suggestions: Provide API to retrieve spelling suggestions. Work started in suggestions branch.

• Detect language of text content and use language specific stemmer.

• Provide an RDF adapter for RDF clients to query beagle. Partially completed in rdf branch.

• Currently Filters work on both Streams and Files. Change the Filter API so that they mostly work on Streams. This will allow them to filter embedded (archived or in attachment) files without extracting them to temporary files.

• Use 7-Zip SDK for LZMA filtering. 7-Zip SDK is LGPL and has a native C# library. Partially done. There is now a filter for lzma-compressed man pages.

• Opening the results from the Beagle Webinterface currently requires making some complex changes in firefox config. Write a simple extension which handles opening of results. It will also make it easier to open custom applications for certain kinds of results.

• Currently Filters have their mimetype and extension associations hardcoded in the source. Move the association to user configurable files (probably XML files).

• Backends know best how to open results returned from them. Hence, they should also return the information with the results so that frontends dont have to write long case analysis or guess how to open search results.

• Keep a list of files beagle was unable to filter for the user to view later. Pretty simple to do since beagle already writes an error string in its log file.

• Move PropertyKeywordMapping to an external xml file. This is dependent on the property name changes.

• Add a documentation tile for beagle-search.

• Add suggestions from this thread about better searching in beagle-search.

• Shortcut keys for beagle-search now have to be manually typed in into beagle-settings. Some generic gnome keybinding should be used. See http://bugzilla.gnome.org/show_bug.cgi?id=410759 and http://bugzilla.gnome.org/show_bug.cgi?id=349257

• Tracker claims to have some new super-awesome system for being the most unobtrusive indexing system around, check out how valid these claims are, and if there are some new ideas, implement them in Beagle.

• If there are no snippets, then return the first 2 or 3 lines. An ultimate goal would be to return a autogenerated text-summary of the content.

• There are API and examples on how to remove an item from the index. Provide a GUI way to remove items from beagle-search. The main difficulty seems to be deciding how to inform the user that when removed like this, the item will be re-added upon re-indexing (e.g. if the file is changed or during a recrawl); for completely ignoring this item it should be added to the ignore list.

• There are several new tools that are now shipped with beagle but does not have any man pages. Also for some of the existing tools, their man pages need to be updated with the current options and description.

• Merge beagel-xesam-adaptor in main project. Some applications need Xesam support. It would be more comfortable for users to install apps like GNOME-DO XSearch.

Last updated after beagle-0.3.7.

- 07:22, 8 July 2008 (EDT)

--Kkubasik 18:41, 3 March 2008 (EST)