'<' and '>' in `apply_to` filters are used the opposite way of their usual meaning when used with 'u'

[uschebit]

When defining filters for the apply_to command, '<' and '>' seem to be used with their meanings reversed on the modification time 'u'.

From README.md:

# removes all clients that have not been updated in last 2 days
at u<2d,t=1 remove_contact
# gives traces to repeaters that have been updated in the last 24h and are direct
at t=2,u>1d,d cn trace

I'd expect the first example to be at u>2d,t=1 remove_contact, because if the clients have not been updated within the last 2 Days, their modification time u is greater than (>) 2d.

Same with the second example: Repeaters that have been updated in the last 24h should have u<1d (u less than 1d).

Is there a specific reason for this rather unintuitive use of '<' and '>' that I'm missing?

[fdlamotte]

I'm aware of this but the d and h postfixes have been added as a convenient shortcut but the argument is really a timestamp (which is compared with last activity timestamp of the node)

I didn't want to change the direction of the symbol when using "ago shortcut" and timestamp so that's why it seems reversed when using shortcut. at u>2d should be read "timestamp > 2d ago timestamp", it's the way it has been written ... Maybe the documentation is not explicit enough