libarray

libarray is the basic structure for a time series storage. Therefor
we have special functions that depend on a timestamp. 

Who need it ?
Lets assume you read a few MB from one or more sensors. every data has a
time stamp and a value. You store everything in a relational database.
unfortunaly RDMS are not very well suited for TS data. You need a lot
work to get and the keep them running. A flat file system is much better,
requieres less attention and takes advantage of the natural order.

How data are stored (struct metadata, see: man aopen)
To store data you need to specifiy a few data. every entry is a 
string and that is translated into filename. using special characters,
Spaces and so is clearly NOT clever.

1. Messnetz
You run a sensor array ? fine, lets call it SIL (sensors in large)
That is the name of the toplevel directory where the data are stored.

2. Methode
You may run differend measurement methodes inside the Messnetz. The idea
is that these is done *inside* the sensor. perhaps you have 
different types of detectors like NaI or HPGe.

3. Kennung
Is something like a stationID. It can be channel_1 or 0124747, but
this is the filename so it has to be *uniq*.


4. Zeit
Normaly invinsible to the user, it describes an additional order.
Because it would produce to large files, the data are broken down to
daily series so we have a Zeit entry also. the entry is translated into
a YYYY/MM/DD directory structure. So you can run for years only resticted
by disk space. The access is fast as on day one.

Why should i use that librarry ?
It allows a fast and simple access to the time series. you can update and
insert but not delete. deleting raw data is bad science.

Why should i use ASCII data ?
We could have used a binary data format but access with grep,sort would be
more or less impossible. Basicly there is no restiction you can change that.

Does it always store in the local dir ?
no, you can use TSDIR to influence that.


Example 1:

You run a a Messnetz called SIL. The methode is NaI and the station ID is
app_1. Now ist 2004-01-01 06:01:00. store all the data in the meta structure. 
Your data should look like that:

2004-01-01 06:01:00|all the data you like to store.

please note that the time has to be stored in the metastructure AND the file.
now call append(). Thats it. most data in the meta structure are constant.


Example 2:

There is file with test data. ./translate test.data will create an example.
it will create a dir SIL/16-kanal-zaehler/2004 with 04 and 05 data.
the file is P3_AlphaRnTn.
							/* start */   /* stop */
./tsdump -N SIL -M 16-kanal-zaehler -K  P3_AlphaRnTn -V 2004-04-30 -B 2004-05-01
will create a dump to the screen.

I used it to feed gnuplot with the data.


INSTALL

There is no install script now because its a first release. 
look how to use the stuff befor you make a mistake.
