2016-09-26 Notes on Release 4.0.0
===============================================================================

New algebra modules:
--------------------

* Distributed2Algebra
     Provides distributed query processing. One Secondo system called the 
     master controls many Secondo instances called workers running on 
     different computers/cores. So-called distributed arrays are provided 
     whose field are stored in a distributed manner on the workers. 
     Operations are applied in parallel to all fields of a distributed array. 
     Fields can be of any available Secondo type, in particular, relations or 
     nested relations.
     
* MainMemoryAlgebra
     This algebra allows one to load existing Secondo objects (e.g. 
     relations) or query results into main memory and keep them there for 
     further querying. The algebra controls available and used amounts of 
     memory and maintain a catalog of objects loaded. AVL-trees, R-trees, and 
     M-trees are available for main-memory indexing.
     
* DBScanAlgebra, OpticsAlgebra
     These two algebras provide the well-known algorithms for density-based 
     clustering, DBScan, and OPTICS. Clustering may depend on geometric 
     distances or arbitrary user-defined similarity functions.
     
* TrajectorySimilarityAlgebra
     Provides operations to measure similarity of trajectories in different 
     ways, to support different applications.
     
* GuideAlgebra
     A comprehensive algebra for the purpose of teaching all aspects of 
     algebra implementation in Secondo.
     
* FileIndexAlgebra
     Allows one to create B-tree or R-tree indices on top of binary files 
     storing relations. Useful in a distributed environment, as index files 
     can easily be moved to other computers.
     
* CassandraAlgebra
     Supports the coupling of Secondo with Cassandra. See the work on 
     Distributed Secondo 
     (http://dna.fernuni-hagen.de/Secondo.html/DSecondo/DSECONDO-Website/index.
     html).
     


Optimizer
---------

The optimizer is now available in a basic version (directory OptimizerBasic) 
and the current version (directory Optimizer). The basic version reflects the 
state of optimizer implementation from about 2006. It is provided to 
facilitate understanding of the basic concepts of the Secondo optimizer 
(omitting all the extensions of the current version to keep it simple).

On the other hand, the basic algorithm for conjunctive query optimization has 
been extended to support the management of properties for intermediate 
results. This could be the order of a stream of tuples, or the distribution 
criteria for a distributed intermediate result. In both directories, the old 
query optimization algorithm is still available in the file optimizer.pl, the 
new one in file optimizerNewProperties.pl.

New textual user interfaces have been introduced that allow one to mix 
queries at the SQL level with queries in Secondo executable language. They are:

  * SecondoPLTTY
  * SecondoPLTTYCS
  * SecondoPLTTYNT

The first one is a stand-alone interface comparable to SecondoTTYBDB, the 
second the client-server version comparable to SecondoTTYCS, and the third 
one the stand-alone version running without transactions, like SecondoTTYNT.
All three versions can run scripts combining executable language and SQL 
commands.

In contrast to SecondoPL, which requires Prolog terms as input, these 
interfaces as well as the Javagui allow one to enter SQL queries and commands 
in a more natural syntax, as it is NOT necesssary any more to use lower case 
names. Instead, all names of Secondo objects can be written in the same way 
as in the kernel, including relation and attribute names.