Particularly as the TWCs eye high sample collocated data there will be a huge penalty in my clunky way to hold waveform buffers in HDF5. I've investigated the Boost Circular Buffer some and I believe it will be the most expedient / efficient / thread-safe strategy towards holding continuous waveform buffers up to a certain period of time and ensuring gaps are filled with easy to identify NaN's. In my opinion this is an Earthworm deficiency - it really should have a scalable RAM waveserver. NEIC has a Windows based one which doesn't do us much good. It'd be great if someone wants to use this opportunity to bring a RAM waveserver to non-Windows machines so that we may eject this part of the G-FAST code.
Particularly as the TWCs eye high sample collocated data there will be a huge penalty in my clunky way to hold waveform buffers in HDF5. I've investigated the Boost Circular Buffer some and I believe it will be the most expedient / efficient / thread-safe strategy towards holding continuous waveform buffers up to a certain period of time and ensuring gaps are filled with easy to identify NaN's. In my opinion this is an Earthworm deficiency - it really should have a scalable RAM waveserver. NEIC has a Windows based one which doesn't do us much good. It'd be great if someone wants to use this opportunity to bring a RAM waveserver to non-Windows machines so that we may eject this part of the G-FAST code.