Following a few coverity scans, several inconsistencies / potential bugs have been fixed. Since my commit to picojson got merged, the newest version of picojson has been taken.

Some of theOSDH file names have changed. Some of the fields are not available anymore. Code cleanup in DBO.

I forgot to change the reference datasets after correcting the Unventillated_T filter... This is now done!

still moving forward with the DBO plugin, picojson has been edited to fix the warnings (few changes), fixing a misplaced space in SetCompilerOptions

In oshd, the stations' names are followed by the network they belong to, separated by a space. This messes up the log analysis scripts of Snowpack, so now the names are cleaned up in order to remove both networks names and spaces.

The Nakamura and Huwald algorithms in Unventilated_T were not properly implemented... This has been fixed and they should now really follow their respective papers. The DBO plugin has been a little bit cleaned up, waiting for new developments of the REST API.

Added a low level json function (getStrings) and improved goToJSONPath() so it supports arrays in the middle of the path

The "NOW±xxx" time specification has been improved (more flexible) and the documentation expanded. Otherwise, code cleanup (code factorization).

Added a "NONE" filter so it is possible to overwrite filters when importing another ini file. The documentation has been expanded accordingly.

code restructuring in DBO to make it more similar to other plugins. Implemented a proper support for recursive METEOPATH in OshdIO (only the most recent files will be used for each time step, according to their run_date).

Another tiny fix related to end of lines: if too many "empty" lines are found, this could be a problem with EOL encoding. This now throws an exception (after 1000 empty lines).

Added the structure for the upcoming DBO plugin, fixed an invalid copyright in GSNIO (the old plugin was ©epfl, the new one is ©slf), added a note to the oshd plugin.

For virtual stations, a time offset parameter has been added to the resampling specification. The documentation has been expanded and improved. Otherwise, code cleanup.

Finally!! The virtual stations seem to be in proper order! Now the spatial interpolations are only called every vstations_refresh_rate and otherwise the data is temporally resampled.

Finally, the proper handling of virtual stations is taking shape: using two tsmanagers, it is possible to temporally resample the data between spatially interpolated time steps. The strategy to decide which timesteps should be temporally resampled is still buggy, this will have to be improved (currently, it resamples and then spatially interpolates at time when it should resample the spatial interpolations instead).

The Buffer class has been expanded with a new push() call and the documentation moved around in IOUtils

A new option has been added to smet, to skip the plot_xxx headers (default: keep them). The virtual stations were not providing a proper call to only get the metadata, this is now available.

The Date::mod() method had some serious rounding errors that could accumulate. This was degrading the precision. The code has been replaced bya more robust implementation that should offer seconds precision (which is what the class offers).

A bug has been found in IOUtils::seek that lead to returning npos when the vector only contained 1 element (and at the requested timestamp). The OSHD plugin has been brought up to date with the latest content and naming of the oshd data files.

The data_converter code has been cleaned up and made more similar to the snowpack operational "data_qa" tool. An invalid check (< when <= was needed) has been fixed in Meteo1DInterpolator (this was leading to segfaults when a buffer only contained 1 data point).

Code cleanup, documentation improvement (link to WFJ dataset), convenience method in StationData (getAltitude).

Fixed a bug in data_converter than was leading to segfault when a station ID was duplicated in the io.ini. Code cleanup in ImisIO (since I first looked there for the bug!).

A tricky bug has been found (by Leonardo, once again!) when accumulating precipitation beyond the original data set: the accumulation was creating "random" precipitation by accumulating data that did not exist. This has been found and fixed (a few conditions were missing to ensure that no etrapolations would be performed when summing partial contributions).

A tricky bug was found (thanks Leonardo for giving me a test case showing the bug!) when copying data from the buffer back into a vector: the "insert" call from the STL does the copy in the range [start, end[ so we were always missing the last element. Most of the time, this was not an issue since we had such a large buffer, but with small buffers, one could see that the last data point was missing... So now, the "end" iterator is always "last_required+1".

The exclude / keep files were not properly handled in order to be case insensitive. This has been fixed in order to be consistent with the KEEP/EXCLUDE statements. The default buffer centering has been changed to 5% (more logical for the usual use cases).

Necessary adjustments in order to be able to detect Oracle OCCI on Windows. But since OCCI's installation is mostly broken / messy, this might not be the end of the story!