F2python 1.5.0

The F2 DBMS written in Python. Needs ZODB and ZEO to run.

## F2 DBMS in Python

### version 1.5.0 - Th. Estier - 2017.

F2 is an object-oriented Database Management System (DBMS) specially built to support
schema evolution. Its main purpose is to support persistent data for which the initial
schema is very prone to changes. Its underlying design choices are based on reflexivity
(no separation between meta-levels and objects, no DDL) and transposed storage, also
called _column oriented_ (F2 tables are external views).

### Obtaining F2

If you can read this file, you probably already managed to obtain the distribution file,
or alternatively you obtained the sources from the git repository. If you downloaded an
F2 distribution file, then you will have to
unfold it in some appropriate folder of yours, by a command similar to:

$ tar zxvf F2-1.4.1.tar.gz

### Installing F2:

The first step is to fine tune your local python installation.
First install ZODB and ZEO with the following command:

$ easy_install ZODB3

> If the **easy_install** command doesn't work for you, you have to extend your python
installation with the python [setuptools](https://pypi.python.org/pypi/setuptools).

Now type the following command:

$ python setup.py install

### Testing your install:

Type the following command:

$ python -c 'import F2'

If this command returns without any effect or message, then your installation is
a success. If it displays any form of error message including _ImportError_ then
something failed during the preceding steps.

### Testing F2 without server (DB in a simple file):

Type the following command:

$ python bootf2.py file:root.db

You should normally see a succession of lines commenting the ongoing bootstrap of
a F2 kernel. The resulting database will be in the current directory under name 'root.db'.

### Testing F2 with a client/server configuration:

1. verify that port 8080 is not already used on your machine, this is the port
on which your F2 server will listen to clients. If this is the case, replace '8080'
by some other appropriate port number in the file 'start_f2_server' (this is a
simple text file that you can edit by any standard editor).

2. create a local subdirectory called 'var' in the current directory. The
server will store data files, logs and other control files in it.

3. launch your server with the following command:

$ ./start_f2_server

4. bootstrap the server database with the following command:

$ python bootf2.py rpc:127.0.0.1:8080

You will of course replace '8080' by the port number you choosed in point 1.
if you changed it. The **127.0.0.1** address is the standard IP address designing
your own machine. A client program running on another machine would use instead
the IP adress you have indicated to join your machine. This supposes you opened
the port 8080 to other machines. If this is the case, BEWARE because anyone can
use your F2 database while your server is running. When in doubt, ask your local
network administrator.

5. To stop your F2 server use:

$ ./stop_f2_server

### Continuing with your F2 DBMS in python:

From now on, if you want you can read the documentation (in directory ./doc) to
continue your discovery of F2. Fasten seat belts.


Cheers,
Th. Estier - december 2016

--------------------------------------------------------------
### RELEASE NOTES:
##### 1.5 big change: adaptation to python 3.
* tested with python 3.5,
* should be still backward compatible with python 2.

##### 1.4 introduced possibility to prefix a class by the name of its database:
* f2.Person is OK if you have only one such class in your namespace,
* f2.HumanRessource.Person is required if you want if you have 2
classes named "Person" in two different databases.
* fixed bootstrap process for class Database()

##### 1.3 distribution includes now:
* new documentation written in TeX has new schema of Kernel,
* includes f2_rest.py : a small web publisher of a F2 content.

##### 1.1 essentially introduces since version 1.0:
* OID's are now unique to a full F2 database, and never reused. OID's creation
uniqueness is garanteed even in client/server context (using ZEO), it is
both thread safe and transaction compatible: even if a transaction is
rollbacked, created OID's are never recycled.
This has strong implications (simplifications) on restructuration mechanisms.

* some convenient new methods in base python classes,

* bootstrap process is now integrated in the F2 package (instead of a
separated program), this simplifies usage of F2 from inside
application servers (like Radixa or Zope): a program creating a
new F2 database can launch the bootstrap process "a la volée",

* some bugs corrected,

* documentation grew up significantly.