Documentation update (Waveform/WaveformIn)

[aswyatt]

Could the documentation be updated to include WaveformIn - it took me a little while (evening through the code) to find out is an alias for Waveform. I presume it's to keep consistency with the other input types, and the Waveform type is for backwards compatibility. I personally will use WaveformIn since it makes the code more consistent and obvious. It was also a little for confusing as to why Waveform breaks the mould.

[Araneidae]

Yes, this is a good point, and it would make sense to update the documentation.

On the whole the Waveform name is indeed for backwards compatibility, but the reason it is that way is that there is a fundamental difference between the waveform record type and other record types (eg ai and ao). Specifically, EPICS has dedicated and separate record types for "input" and "output" records (though the direction is terribly confusing, as ai records are for outputs from the IOC, and ao records are for control inputs to the IOC!)

However, there is no such distinction for the waveform record, instead it is up to the "device support layer" to decide in which direction the data flows. In particular, when processing a waveform record, the record processing function can choose whether to read a freshly updated waveform from the record (treating as an "out" record) or to update the waveform in the record (making it an "in" record). Indeed in principle a waveform record can dynamically switch between these two behaviours (though pythonSoftIoc does not support this)!

[Araneidae]

Fixed in 5f02a84