The first version of an ARPSIO plugin has been written, largely based on the code of LegacyIO. This plugin is mostly non-fonctional (besides reading DEM) because most of the concepts of ARPS files are foreign to MeteoIO... Finishing it will require carreful thinking and brilliant ideas! (the biggest issues have to do with the irregular grids that are a usual feature of ARPS files).

A warning has been silenced until we finish the implementation of the USER interpolation algorithm (reading user provided grids).

Some commented out code has been removed. The check for the projection of the stations compared to the projection used for the DEM has been re-activated (it MUST be active!). The first implementation of a custom grid "interpolation" method has been done (issue 50) but can not be used: we need access to an IOHandler in order to properly read grids provided by the user. Otherwise, the rest of the code for it has been written and tested.

Then, the remaining namespace issues with doxygen have been fixed, by enclosing the full class implementation in the namespace (as seen for the libstdc++). This removes all these irritating issues. But the popc version has been broken by the change (as seen in LegacyIO).

When using A3DIO for writing, the files are now correctly split by years. The OUTPATH user parameter was previously not used, this has been fixed.

In IMISIO, the query size limitation has been removed. 

A data format converter using MeteoIO has been added to the examples.

METEO = SNOWPACK
NROFSTATIONS = 1
METAFILE = input/IMIS_Extracted_Info.txt
STATION1   = DAV1
METEOFILE1 = /home/thoems/Desktop/snowpack_ref/validation/trunk/snowpack/mst2006/MST06_RR.inp

The parameters NROFSTATIONS, METAFILE, STATION# and METEOFILE# are obligatory.

A little bug in GEOtopIO lead to the copying of the tss value to tsg.
The MeteoData::HS snow height is internally represented in meters.

The A3DIO plugin now has a writeMeteo() method. It writes all timesteps, even if truncating the timestamp to the hour. It also assumes that all stations have synchronized timesteps (ie: reading the timesteps for the first station is enough). It can also not handle metadata that change over time and it does not split 2D files per years...

Some uncommitted fixes to the documentation are now committed. Hopefully all the namespace issues (ie: broken links) are now fixed!

Removed the STATION key from the io.ini. the readStationData function of the plugins relies on the METEO key for choosing the data source.

The STATION keyword in the io.ini tells MeteoIO where to look for station meta data. Mostly this will be exactly the same plugin as for the METEO keyword.

Added the DEBUG_ARITHM option to the CMake build system (can be toggled with ON and OFF: default ON). 

Got rid of some verbose output in the RateFilter.

Changed the filter buffer size from 2 to approx 10 elements, that should allow for the deleting of certain points within vecFilteredM: changes to the resampling process and the accumulation process were necessary

Small namespace fix for Coords.cc (for doxygen)

Most of the doxygen issues have been fixed. A few are still left (and we lost the inheritance structure in doxygen).

A trivial formatting issue has been fixed for SNIO (consistently writing "0.0" when no data is available for non mandatory fields).

The IMISIO plugin now reads the database's name, user, password from the configuration file. The SNOWPACK plugin has seen some improvements in the formatting of its output file. The code exaples have been simplified.

The new ImisIO plugin with a workaround SQL query to deal with the duplicate records problem in view ams.v_amsio

The POPC compilation issue has been fixed (work around a POPC bug). These work arounds have been tagged as "HACK for POPC" in DEMObject, Date, Grid2/3DObject, IOHandler.ph, LegacyIO.ph and marshall_meteoio.cc.

The SNIO plugin that was forgotten at the last commit is now here and has been improved so that it correctly formats its output (it does not read yet).

Cosmetic changes: documentation in MeteoData, comment in ARCIO. Plus more variables are read by BormaIO.

Changed the interface of MeteoData, the setter function now can be invoked like setData(MeteoData::TA, 280.7);

the old setter function and the old constructor have been removed and the changes were propagated into the plugins

Cosmetic fix in Date.h and some namespace additions for the POPC version (.ph and marshalling). But the marshalling is not working yet...

Adapted the Makefile and the meteoio_demo program to use the renamed Date object and the new namespace mio

Changed the name of the Date_IO class to Date. In case of namespace conflicts, just use mio::Date to reference the MeteoIO Date class.

The doxygen documentation of the filters has been slightly tweaked and the wind averaging filter has been fixed (it was not modulo 360 and an unnecessary sign change has been removed) and verified (closing issue 51). The offending Undefined in Date_IO has been replaced by undefined (for naming consistency).

Two public methods have been added to Date_IO (getJulianDayNumber and isLeapYear). The documentation has been slightly improved. A helper function has been added to IOUtils: getTimeZoneParameters. It will be used by the plugins in order to correctly set the timezone of their data.

The Median Absolute Deviation filter has been implemented, documented and checked (in FilterAlgorithms).

A bug has been fixed in the rate filter, a better behavior has been implemented for the Date_IO class (once a timezone has been set, either by the constructor or by an explicit call, all setDate() calls that don't specify a timezone will be considered to be in the pre-set timezone), the class has been documented and the example has been improved.

The Date_IO class has been deeply rewriten. The conversion methods have been checked (with references added), the class now uses real julian date (internally as a reference) and can generate a variety of outputs (julian, modified julian, truncated julian, unix, excel). It also works properly with time zones and a simplified version of dayligh saving time (it assumes that DST always introduces a +1 hour shift, which should always be true but on one AUstralian Island: Lord Howe Island). Some now invalid calls have been fixed in plugins (the short setDate(yy,mm,dd) call is now longer supported but migth be misinterpreted by the compiler for another call, despite types mismatch).

A new example code has been added: time.cc