Nucular Fielded Full Text Indexing / News: Recent posts

The 0.5 nucular release includes boolean query syntax so you can search for dogs or cats but not snails more easily "(dogs | cats) ~snails".

A thread on the whoosh mailing list found that Nucular was at least 4x faster than Whoosh.
http://groups.google.com/group/whoosh/browse_thread/thread/a2d7b7d2243b1c49

Rene Maurer pointed out problems with unicode handling. I think this is fixed in the development version, available via mercurial at http://aaron.oirt.rutgers.edu/cgi-bin/nucularRepo.cgi -- unless it turns out to be wrong the fix will be included in the next file release.

The readOnly archive option was broken in 0.4. Fixed in 0.4.1. Sorry!

There is a new 0.4 release:

This release adds a simple "sql like" wrapper
which makes Nucular easier to use for some
purposes. It also fixes a number of bugs.

The file storage format for this release is
not compatible with older formats (the change
was needed to fix a bug).

On Windows XP I had to delete the old
package manually from the Python library
before the install for the new package
would work properly.

Nucular beta 3 just released!

Nucular now supports WIN32. Current releases
of Nucular abstract the file system in order
to emulate file system feature missing
on NT file systems which prevented older
versions from running correctly on Windows NT
based systems.

Proximity search added: Current versions of
Nucular allow queries to search for a sequence
of words near eachother separated by
no more than a specified number of other words.... read more

Someone suggested that nucular would be faster with Python garbage collection disabled. This is true! For index builds I see a 30% speed up after adding gc.disable(). For retrievals on a "warm" index I also see speed ups of about 10% under some circumstances. Since large index builds can take hours the savings could be significant. Please take note. Future releases will include logic to automatically turn off garbage collections during expensive operations. For now you might want to add it yourself. FYI!

Quick clarification: you can *query* a nucular index on windows. I haven't tested it but you might also be able to *build* a nucular index on windows. What you can't do is build and query *at the same time* (as yet) because that requires Posix file system semantics. If someone wants it badly enough I think there are ways to support full functionality on Windows. Let me know.

I just put out a new release of NUCULAR and declared the project at beta because it seems to work for people and the code is stabilizing. The release fixes some minor problems, tweaks the core data structures and adds a handful of cute features.

The first release of nucular full text fielded indexing is out. I just noticed I forgot to document the session.cleanUp() method, which is pretty important (that's why all the test and demos use it). Other than that the release is absolutely perfect and bug free :). I hope you like!