all(::TimeSeries)?

[dourouc05]

Following a recent trend of addition of methods (like #338), would something like all()be interesting? I am currently comparing two time series, and a simple == generates a warning:

WARNING: ==(ta::TimeArray, args...) is deprecated, use ==.(ta, args...) instead.

I have thus switched to ==., and I have a time series of Booleans, instead of a single Boolean. I would like all(ta) to return all(ta.value) when that makes sense (I am only thinking of having one Boolean column and no other column). Then, any() could be implemented similarly (and maybe others).

This is more a request for comments; I can provide an implementation if you think this is worthwhile.

[iblislin]

Following a recent trend of addition of methods (like #338), would something like all() be interesting?

I agree with that we should support more reduction functions like all and any,
but the next problem is the return type. Should it be a scalar, be a TimeArray or depends?

[dourouc05]

When there is only one column, I think the answer is quite obvious (take the all of the values of this single column, meaning a single Bool). However, when there are multiple columns (only Booleans), it might make sense to do it line by line (i.e. return a new time series with just one column of Bool).

[iblislin]

First, for your need, I think we need to revoke the deprecation warning of == and redefine its meaning as 'Does two TimeArray share exactly same fields?`

Second, about all, any and etc, I will prefer to return a TImeArray like other reduction function (sum, mean) does. Since TimeArray is a tabular data structure and it has two dimesion,
in order to do reduction, you need to specificy the dimension (the default is along the columnes).
That's what #338 does: e.g. sum give you a TImeArray with latest timestamp.
I want to keep consistent on all and any.
Futhermore, I will expect the reduction function shares same result with moving(..., full_length)
e.g. moving(sum, cl, length(cl)) equals to sum(cl).

[iblislin]

you might want to try out #359