flexmeasures delete prognoses fails with TypeError

[Ragnar-the-mighty]

I am on v0.21
When I call
flexmeasures delete prognoses --sensor 267

I get this error
File "/opt/anaconda3/envs/flexmeasures-venv/lib/python3.10/site-packages/flexmeasures/cli/data_delete.py", line 224, in delete_prognoses
depopulate_prognoses(db, sensor_id)
File "/opt/anaconda3/envs/flexmeasures-venv/lib/python3.10/site-packages/flexmeasures/data/transactional.py", line 38, in wrap
db_function(*args, **kwargs)
File "/opt/anaconda3/envs/flexmeasures-venv/lib/python3.10/site-packages/flexmeasures/data/scripts/data_gen.py", line 351, in depopulate_prognoses
click.echo("Deleted %d forecasts (ex-ante beliefs)" % num_forecasts_deleted)
TypeError: %d format: a real number is required, not CursorResult

[nhoening]

Indeed, above that line we have num_forecasts_deleted = db.session.execute(query). That used to work, but I suppose that after we moved to SQLAlchemy 2, it no longer does (actually there are a few places in data_gen.py which have this same problem).

This part of the SQLAlchemy documentation suggests to use .rowcount, so this should be better:

deletion_result = db.session.execute(query)
num_forecasts_deleted = deletion_result.rowcount

Are you able to try out if this fixes your problem?

[Ragnar-the-mighty]

The code change removed the error but the result I got was
Deleted 0 forecasts (ex-ante beliefs)

I do have forecast so this is not the behavior I expected.

[nhoening]

Maybe the previous run had already deleted them?

[Ragnar-the-mighty]

No I am simulating back in time so my horizon is negative. I think that is why nothing got deleted.
I am still learning about the time capabilities of sensors.

[nhoening]

Ah, yes that is different, then they count as measurements.

When you create beliefs, you can set the horizon which fits the assumption of the simulation. We are using FlexMeasures to simulate internally, as well.

For deleting whatever beliefs of a sensor (or even per asset), try flexmeasures delete beliefs --help

However, that code might also exhibit the same problem you point out in this issue (namely in flexmeasures/cli/data_delete.py), e.g. here:

num_beliefs_up_for_deletion = db.session.scalar(select(func.count()).select_from(q))

Do you want to add the fix you tested into a PR? That would be very welcome (and make you a FlexMeasures contributor :)) - but I can also address it myself. Thanks for pointing it out in any case!

[Ragnar-the-mighty]

Let me test a little more then I try to address the PR.
I make a comment if I need some help fixing the PR.