directory-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Emmanuel L├ęcharny <>
Subject Next steps...
Date Wed, 21 Aug 2013 08:08:32 GMT
Hi guys,

as I'm going to be off for one week, and as we just cuut a new reelase,
I'd like to share some random thoughts about the server, the API,
Mavibot and some other aspects of our project, where we are now, and
what remains to be done in thenear future...

First, I think we are slowly but consistently closing the gap to a
2.0-RC1. The last release was mainly about fixing some bugs, and the
next one is probably the first RC. What remains to be done in order to
cut this RC1 ?

o Make Mavibot a first class partition (althouh still experimental). It
should be possible for anyone to configure the server so that the
partition uses either JDBM or Mavibot, without having to use any weird
JVM options. This is a work in progress, Kiran already pushed some code
that makes it possible to declare a Mavibot partition.
o Avoid the cloning of entries we are returning
o Check the bug list and fix the low hanging fruits

o make the handling of multiple values more performant. We shuld define
a threshold after which we use a subBtree, instead of creating a
subBtree everytime
o improve the performance of writes : we are updating the BTree headers
and teh RM headers after each phisical page update, whihc is adding
ahuge penalty
o complete the bulk load implementation

Not too much in this area. We are close to completion, at least for 1.0.
The doco is still really poor...

o Add the support of ADS 2.0-RC1
o Add support of Mavibot partition

That's all for the short term tasks, as I can see them. I may have
missed some, so please complete this list.

Otherwise, in a middle term vision, we have many areas where we can
improve the server and the other projects :

o switch to MINA 3
o review the way we handle writes : we should wait for the messageSent
event before writing a new message, otherwise we cumulate them in memory
o review the encoding part : using a pre-allocated buffer stored in a
ThreadLocal storage would save us some CPU cycle, as we could get rid of
the allocation of a buffer
o review the ACI subsystem
o fix the administrative model, which is not complete
o add the delta-syncrepl protocol for the MMR subsystem
o implement the missing LDAP entities (NameForm, DCR, DSR, MRU...)
o make the server OSGi compliant
o ditch JDBM as a backend
o get rid of most of the locks used for JDBM

o implement the missing LDAP entities (NameForm, DCR, DSR, MRU...)
o implement a better support for some other server schema (OpenLDAP,
OpenDJ, AD, etc)

o implement the transaction support
o manage MVCC free pages
o add a checker and a fixer after a crash

o go for a full web based Studio
o add configuration support for other servers

This list is just limited, again, please add the items you think it's
important to have.

Comments very welcome !

Emmanuel L├ęcharny 

View raw message