Alessio Buccino

> > Removed IblStreamingRecordingExtractor in favor of IblRecordingExtractor (@CodyCBakerPhD any changes on neuroconv required?) > > Nothing was ever added to core NeuroConv since it was such a specific usage...

@h-mayorquin ready to merge on my end ;)

> Quick (maybe dumb) question: what about when there are multiple parents? (e.g. `UnitsAggregationSorting`) It returns `None`, but we have other ways to return the parent list (`property sortings/recordings`)

> 1. For the segments we have the weakref mechanism. Would it make sense to have smoething like this for this? Does the context were the other decision was made...

> > 1. For the segments we have the weakref mechanism. Would it make sense to have smoething like this for this? Does the context were the other decision was...

> I have the intuition that we should make a weak reference instance of a hard link. Lets maturate a bit this before merging. I think it doesn't make sense,...

> Since Windows is the biggest culprit for problems is there a test I can run locally to probe this? I don't think there is! IMO let's merge and see...

I agree with @zm711 that `spike_indices` is an "advanced" concept which will make the API over complicated. Fine to use it internally, but at the higher level `compute_random_spikes` or `sample_random_spikes`...

Thanks @manimoh This is great! Could you add a test for it? We have several SpikeGLX test files in the [GIN epy_testing_data](https://gin.g-node.org/NeuralEnsemble/ephy_testing_data) repo, and I'm sure (hope) at least one...

Thanks! @samuelgarcia can you upload to GIN?