Support NXlog with "sublogs" such as connection_status and alarm #138
Conversation
| time-indexing, since the time coord of each time-series is different. We can | ||
| support label-based indexing for this in the future. If additional time-series | ||
| are contained within the NXlog then loading will return a DataGroup of the | ||
| individual time-series (DataArrays). |
There was a problem hiding this comment.
If we had some form of flexible structured dtypes, could this data be merged into a single data array by binning? (e.g. with the main log's time coord)
There was a problem hiding this comment.
I don't think so. What do you have in mind?
There was a problem hiding this comment.
da['time', i] contains the value of the main log in this bin as well as all values of sublogs that fall into this bin.
There was a problem hiding this comment.
Typically the main log has many more values than the sublogs (often by orders of magnitude), so I don't see how that would be useful/feasible?
| for name in self._sublogs: | ||
| for k in list(children): | ||
| if k.startswith(name): | ||
| field = children.pop(k) |
There was a problem hiding this comment.
This modifies the argument children. Please make a copy first.
| for name in self._sublogs: | ||
| # Somewhat arbitrary definition of which fields is the "value" | ||
| value_name = 'severity' if name == 'alarm' else None | ||
| sublogs[name] = self._assemble_sublog(dg, name, value_name=value_name) |
There was a problem hiding this comment.
_assemble_sublog also modifies an argument (dg)
This is the way ESS tries to store additional values at distinct time-stamps. It is apparently being proposed for standardization in NeXus.
It is kind off a mess and not possible to handle nicely (see comments/docstrings), but apparently this cannot be changed, so we have to deal with it somehow.