[Cpu-users] cpu-1.4.2 and feature freeze

[Blake M. <bma...@pu...>]

Hello All,
cpu 1.4.2 was released today and is available at:
http://sourceforge.net/project/showfiles.php?group_id=45948

This release is the last release in the 1.4 series to contain new features.
The only new features that will be added are ones that will not possibly
introduce new bugs into existing functionality. The only changes other then
that will be bug fixes. The reason for this is I'm going to be branching a
1.5 series which will include a rewrite of the API interfaces. In order to
alleviate problems with upstream maintainers and bugs, a feature freeze on
the 1.4 series is needed. A rewrite of the API interfaces is needed in order
to better facilitate people writing other backends. The existing 'API' is
unfortunately fairly useless.

There were three big changes to this release. The first was the move from
synchronous to asynchronous searches, the second was the combination of using
asynchronous searches with a bitvector, and the third was the way the help
works. Comparison between adding a user with
version 1.4.1 and 1.4.2:

# users         RANDOM=?        Version Time
1000            true            1.4.1   0m0.421s
1000            false           1.4.1   3m34.848s

# users         RANDOM=?        Version Time
1000            true            1.4.2   0m0.437s
1000            false           1.4.2   0m0.547s

# users         RANDOM=?        Version Time
5000            true            1.4.1   0m2.448s
5000            false           1.4.1   130m19.123s

# users         RANDOM=?        Version Time
5000            true            1.4.2   0m3.530s
5000            false           1.4.2   0m1.887s

where '# users' was the number of users in the directory (uid 0-#), RANDOM=?
was the setting random in cpu.conf. The rest is self explanatory. The time
difference was significant enough that I thought this warranted a release.

I also overhauled the error reporting mechanism to give users useful feedback
as to what the error was and what can be done to resolve this. The reason for
doing this was two fold. First, to allow for better internationalization
support and the second because most of the OpenLDAP error messages are
terrible.

There were some minor bug fixes as well.


-Blake


-- 
Blake Matheny           "... one of the main causes of the fall of the
bma...@pu...      Roman Empire was that, lacking zero, they had
http://www.mkfifo.net    no way to indicate successful termination of
http://ovmj.org/GNUnet/  their C programs." --Robert Firth