The dataGenerators can now process all parameters (ie including the extra parameters) properly. This makes it possible to create a new parameter by using a COPY followed by *0, +nodata filters and then fill the now empty new parameter with a data generator (very useful to compare generated values with a reference measurement). Moreover the processing overhead should have been reduced by looping over the declared generators instead of over the parameters.

In order to implement an efficient and portable object serialization, it has been decided that all serializable objects would be manually serialized to an std::iostream. Then, the parallelization framework (mpi, popc, etc) can take over this stream and move it to another node. Therefore, the operators "<<" and ">>" are now reserved for serialization (as seems to be standard) and the debug outputs are obtained by printing "object.toString()". This means that all debug code has potentially to be fixed, by replacing "std::cout << object" by "std::cout << object.toString()".

Further improvements for the resampling within the Meteo1DInterpolator: no longer is an element inserted into the filtered_cache and then deleted, instead we have a MeteoData reference handed to the Interpolator representing the MeteoData element to be resampled. This interface change needed to be propagated into the different affected classes. Accumulate resampling currently disabled.

Speedup in the resampling of MeteoIO, instead of always copying a resampling window we are now directly resampling on the filtered_cache. This leads to an enormous perfomance gain (slashing runtime for reading meteo data to fifth), because the cutting of the resampling window was quite costly.

Enhancement: It is now possible to duplicate a certain meteo parameter when reading data with MeteoIO. This feature is especially useful when one parameter needs to be filtered in two different ways (e.g. a running average spanning 5 hours and one spanning 48 hours). The syntax to use in your io.ini:

[Input]
COPY::TA2 = TA

This means that a new meteo parameter called TA2 will be added when reading meteo data from an IO source and that this new parameter will be initialized with the value of MeteoData::TA. Thus there is a copy of TA called TA2. This copy can be now used to set up a specific filter:

[Filters]
TA2::filter1 = mean_avg
TA2::arg1 = right 5 3600

Fixing some unsigned int / size_t conversion issues by using more widely size_t (as it seems to be recommended and done in the STL)

Another commit related to timezones... The Unix date always being in UTC, this has been fixed. The resampling was not using dates in a proper way, this has been fixed. The GSN plugin has also been fixed, it now rounds time to the closest minute (in order to be consistent with our advertised "1 minute time resolution"). A rounding method has been implemented for dates.

The timezone parameter in io.ini is now named TIME_ZONE instead of TZ for clarity. 

Undefined dates should now be fully handled (ie: also in arithmetic).

Finalization of the new filtering infrastructure: speed optimizations (IOManager, MeteoProcessor), clean up

The MeteoFilter object is obsolete, it has been fully replaced with the ProcessingStack class. Added comments to MeteoProcessor.h

New filter system rolled out with currently two filters up and running: min_max and mean_avg (without the soft option, only left or right centering). The new filter system is built upon processing blocks that take a vector<MeteoData> as input and return a processed vector<MeteoData>. The individual filters are stacked in a ProcessingStack where they are invoked sequentially.

Introducing the new IOManager - the user interface to meteoio. This version is fully backwards compatible, but does not incorporate all features of the BufferedIOHandler: the grid reading and writing are currently not supported by this IOManager.

The unnecessary copies of a Config object have been removed from MeteoFilter and MeteoProcessor. Now, cfg is used in the constructor and not kept any longer. Next task: do the same for Meteo1DInterpolator!

The smoothing algorithms have been renammed as Process instead of Filter. The rate and the Tukey filters have been improved so that they correctly handles nodata.

A new resampling algorithm has been introduced: "none". It does just that: nothing. This is an easy way to debug filters by not performing any temporal interpolations.

The getMedian method has been re-implemented using the standard nth_element method for increased performances.

The doxygen documentation has been updated to reflect the latest changes.

MeteoData now has an extra public member variable: StationData meta. It will hold the meta information of the measurment instead of dragging along a second vector<StationData> alongside the vector<MeteoData> everywhere.

This has huge implications: the IOInterface has been rewritten, subsequently changes need to be propagated into all plugins and all modules dealing with meta data.

More debugging help: the "<<" operator has been redefined for MeteoFilter, MeteoProcessor and BufferedIOHandler. Since the Config object is copied all over the place, it is not always printed.

Fixed the behavior of ConfigReader when throwing exceptions: it rethrows so that the error message can be turned off by the receiver of the exception (if simply not printing e.what() when catching the exception). 

The warnings in FilterAlgorithms have been turned off cast to void) since this is an algorithm that has to be re-written anyway. The documentation has been fixed (missing documentation, broken link, etc)

The new filtering and resampling framework. Filtering/Processing and Resampling have been separated from each other, the Resampling is now done with the class Meteo1DInterpolator which utilizes the static algorithms within ResamplingAlgorithms. The filtering remains a task of class MeteoFilter. The calling of the resampling and filtering as well as the data selection is now based in the facade class MeteoProcessor. MeteoProcessor::processData function calls the filtering algorithms, then passes the filtered data to the resampling algorithms and finally runs a "check only" filtering pass.

The filtering system underwent some changes in order to simplify the tasks at hand, the interface has changed.

The RateFilter is currently disfunctional, needs to be reimplemented.