Support:Search v3

DRAFT
The content of this page is a work in progress intended for review.

Please help improve the draft!

Ask questions or make suggestions in the discussion
or add your suggestions directly to this page.

This document refers to the search rewrite of Q4 2009.

Goals

• Improve index time (should be able to index at least once/hour).
  • Move from Sphinx 0.9.8 to 0.9.9.
  • Replace xmlpipe2 with direct SQL access.
• Improve advanced search.
  • Implement new UI. [1]
  • Meaningful options for both knowledge base and forums.
• Improve spelling suggestions.
  • Move from SOUNDEX to Aspell.
    • Dramatically improve suggestions.
    • Will hopefully improve performance.

Timeline

Search v3 is the primary goal of SUMO 1.5. Our current schedule is to code freeze on November 24th and release on December 1st.

Code

The Search v3 branch has been merged into trunk and the branch is closed.

https://svn.mozilla.org/projects/sumo/trunk

The Search v3 code is now up on our normal staging server.

http://support-stage.mozilla.org/

The username/password is "support"/"stage".

Indices

The current sphinx.conf provides the indices, with the indicate attributes:

forum_threads

Returns a threadId (for a thread)

attr name	type	description
forumid	int
author_ord	int	crc32(userName)
status	int	crc32(status)
created	timestamp
last_updated	timestamp
replies	int	number of replies
authors	multi int	crc32(userName) for all posters in the thread

forum_posts

Returns a threadId (for a specific post)

attr name	type	description
threadid	int	threadId of the actual thread (not the post)
author_ord	int	crc32(userName)
status	int	crc32(status)
forumid	int	forum id for the thread
created	timestamp

wiki_pages

Returns a page_id

attr name	type	description
lastmodif	timestamp
locale	int	crc32(lang)
tag	int multi	crc32(tag)
category	int multi	categId