davidkohn88

Hey @sbahman if you want to give it a try an initial version is up. Documentation isn't fully done yet, but something to work with.

Hi! Yes, we currently don't support Windows. I've changed this to an enhancement label, and people can upvote here and we can see how to prioritize based on that. Thanks...

@charislam do you think you could make the updates? Also, we'd probably recommend as a workaround running PG in a Docker container on Windows. Just as a note.

So what you'd like to do, I suppose is be able to do something like: ```SQL SELECT signal_id, time_value, toolkit_experimental.interpolated_average( tws, time_value, '3600 s', LAG_IGNORE_NULLS(tws) OVER (Partition By signal_id ORDER...

> Do i understand you correctly that the methods **LAG_IGNORE_NULLS** and **LEAD_IGNORE_NULLS** are not yet existing ? > Yes, something like your last query would be very useful :-). Yes....

> I have the same problem, My current workaround is to use a 3 steps approach using time_bucket_gapfill first, and then use time_weight on the gapfilled data. > > Example:...

Looking at this again, I think I found a more efficient workaround: ```SQL with t1 as (SELECT time_bucket_gapfill('{resampled_interval} s', ts) AS tsl, signal_id, locf(last(measurement_value,ts)) as locf_value, time_weight('LOCF', ts, measurement_value) AS...

```SQL create or replace function time_bucket_range( bucket_width interval, ts timestamptz) RETURNS tstzrange as $$ select tstzrange(time_bucket(bucket_width, ts), time_bucket(bucket_width, ts + bucket_width), '[)'); $$ LANGUAGE SQL IMMUTABLE PARALLEL SAFE; ``` Should...

Unfortunately, I don't think this is a bug, if there is one occurrence without bounds the time weighted average has to be null, the way it is calculated you need...

Hey @davetapley sorry I missed this, got lost in the holiday shuffle I think. Can you take a look at https://github.com/timescale/timescaledb-toolkit/issues/548 and see if that looks like a similar case?...