Expose gaps in moving window

[daniel-zullo-frequenz]

What's needed?

Implement method to return the maximum data gap in a MovingWindow given timestamp or index ranges representing a window size

Proposed solution

Compute the maximum data gap size iterating through the MovingWindow data gaps (underlying buffer) for the given timestamp or index ranges representing a window size

Use cases

Actors like peak-shaving needs this functionality to avoid implementing it on their ends.

Update

Note the scope of the issue was updated as follows:
The MovingWindow will provide access to a list of all gaps in a defined subset of the moving window. Applications can get the maximum gap or the sum of gaps by using something like max(g.duration for g in gaps) or sum(...), respectively.

[daniel-zullo-frequenz]

fyi @matthias-wende-frequenz @idlir-shkurti-frequenz

Please feel free to amend/correct the issue description accordingly

[cwasicki]

After discussion with @idlir-shkurti-frequenz we change the scope of this issue. The max gap is currently used in peak shaving but is going to be replaced with a sum of gaps soon. Therefore we plan to provide access to a list of all gaps in a defined subset of the moving window. By this current and future requirements can be achieved using something like max(g.duration for g in gaps) or sum(...), respectively.

[llucax]

If there is a PR for this, can you link it to the issue @cwasicki ?

[cwasicki]

Since it's not a breaking change I think we can move it post v1.

[llucax]

@daniel-zullo-frequenz said he thought there was a PR for this, so I guess this is not the case, right?

[cwasicki]

I have some changes locally which I can push and link

[cwasicki]

#643

[cwasicki]

@daniel-zullo-frequenz Could you update title and description to something like Expose gaps in moving window (see #558 (comment))

[daniel-zullo-frequenz]

done

[cwasicki]

Discussed with @idlir-shkurti-frequenz that this can be closed as they can use the new moving window functions

frequenz-sdk-python/src/frequenz/sdk/timeseries/_moving_window.py

Line 357 in a06db70

def count_valid(self) -> int:

and

frequenz-sdk-python/src/frequenz/sdk/timeseries/_moving_window.py

Line 366 in a06db70

def count_covered(self) -> int:

to determine the data quality of the window in peak shaving.