* remove support for headonly keyword when writing H5 files (reading with headonly=True still supported) * change default index to flat index including the trace number as running counter -> traces will not be overwritten by default, even traces with the same metadata -> support for traces without SEED id out of the box (raised an error with default index in version 0.5.0) -> Stream has the same order after writing and subsequent reading (for the default index only) * other pre-defined indices are "flat", "nested" and "xcorr" * allow to use format specifiers {id} (SEED id) and {duration} (in seconds) when creating new indices * remove offset_trc_num keyword, not necessary anymore, because it is preserved as file attribute
0.5.0
* remove support for python versions < 3.5 * support writing of more types for the header (numbers, strings, np.ndarray, obspy.UTCDateTime, and obspy.AttribDict), format specific headers are excluded by default. * add support for '{trc_num}' in the index which inserts the number of the trace
0.4.1
* allow to set libver when creating a new file (default is now 'earliest' for best backwards compatibility)
0.4.0
* add official support for python 3.7 and 3.8
0.3.2
* Stream returned by readh5 is now sorted
0.3.0
* remove future dependency * change order of optional arguments (e.g. move mode argument to the end for read function) * change default for group from 'waveforms/' to '/' and therfor start the index with 'waveforms/'