[Cpu-users] CPU 1.3.99

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

Hello List,
I'm not actually sure if anyone is still on this or not :-) It's been almost a
year since any fresh code has been written, but I refuse to let CPU become
another abandoned free software project! Having said that, this weekend I
tried to do something about it and completely rewrote CPU. It was a mess,
and not something that I would have wanted to contribute to. The 'new' version
is currently only available via cvs, under the directory 'cpu'. The TODO file
has been included below. Please, if you can donate time or energy to help out,
including testing, please do. Thanks.

------------------------------------------------------------------------------
There is a lot to do in order to get CPU to a place where I can put it to rest
for good. I decided to rewrite all of the code, which while this was a good
idea, doing it in a weekend may not have been. As a result, there are a few
bugs.

I rewrote the system so that it would be more attractive for someone to pickup
and continue to maintain. If you are interested (or will be after I finish
with the fixes below) please contact me.

Here are some of the features that CPU 1.3.99 has that previous versions did
not, and some of the features that are still missing:

New Features
* Code is extremely modular, meaning people can write back-ends easily for any
  kind of system by just exporting CPU_init in their code. This also makes it
  easy for people to include code already written from other projects.
* I'm not so embarrassed to have written it
* New configuration file format allows multiple authentication methods to be
  used
* Modular design allows you to specify which back-end to administer through a
  command line switch which will then load the appropriate module. All command
  line options are specific to the back-end you are administering
* Some API functionality to encourage people to write some new back-ends.
* Rewrote build system
* You can now choose whether to do a serial or random search for a new UID or
  GID as opposed to just random
* We now strictly follow the RFC 2307 standards. However, CPU is still
  flexible enough that you can use other attributes
* You can now decide whether to use uid or cn for your dn

Things that don't work yet (that used to)
* All group functions. Getting them going again will be trivial, all checking
  is already done by the time the appropriate functions get called.
* Usermod function is only half funtional
* New configuration file options aren't documented
* Changes in command line arguments haven't been documented
* No file functionality works (i.e. creating home directories, removing old
  directories, etc)
* Bug checking

Things that people want that I have promised
* TLS/SSL for LDAP
* Support Samba
* XML files for adding weird users

Problems I haven't yet figured out
* Why our included version of getopt() seems to have some serious problems
  with multiple invocations
* What does dmalloc suck so bad? It catches things that make no sense.

Things I'm planning on adding for fun:
* Support administration of flat files!
* NT4 administration
* The stuff that people listed above

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