Re: [Netatalk-devel] afp3-dev testing

[Joerg L. <le...@wu...>]

Libor Vanek:

>> dbd: Access to the CNID database is restricted to a daemon
>> process. afpd processes communicate with the daemon for database reads
>> and updates. If used with Berkeley DB transactions (the default) the
>> probability for database corruption is practically zero, but
>> performance can be slower than with cdb.
>> 
>> 
>> 
> OK - and what branch should I use if I want to test in 
> "nearly-production enviroment" dbd?

There is one CVS branch, called branch-netatalk-CNID-refactored that
contains the initial dbd code. It has the advantage that it uses the
same database format as in the 1.6.xx series, so you can use it
as a drop-in replacement without upgrading the databases. It is
reported to work stable in production for at least two
installations. It is not very well documented, though, apart from a
README file in etc/cnid_dbd, but that should get you going.

Alternatively, branch-netatalk-afp-3x-dev contains dbd as well. There
have been some changes, the most user visible one that Berkeley DB
version >= 4 is now required and the changed database format. I can
help you on how to upgrade if you want give it a go. You can of
course also start from scratch (empty database). Since that branch
will eventually morph into 1.7 using it in near production will yield
important test results.  branch-netatalk-CNID-refactored will not be
developed any further.

>> As far as I know (but Bjoern Fernhomberg is the authority on that)
>> not. The soon-to-be alpha CVS branch branch-netatalk-afp-3x-dev should
>> be able to do this.
>> 
>> 
> OK - what needs to be changed in this? IIRC should be enought to change 
> function for Mac to Unix (and back) filename conversion - or not? I can 
> have somebody to code it when I'll know what functions change :)

I'll really have to defer to Bjoern on that, he'll be able to answer
that much better than I can.

regards,

-- 

Joerg Lenneis         le...@wu...