12:00 <bfiedler> while looking at the bug isaac found in my PR I noticed another interesting behaviour

12:00 <bfiedler> when borders are nonzero, repeated application of set-position 0 0 causes the window to shrink by border_size each time

12:02 <bfiedler> I think it makes sense for set-geometry to set the size of the window's view, ignoring the border size. I'm not so sure about set-position, because intuitively I would expect set-position to be idempotent, which it is not

12:02 <bfiedler> (it will never be idempotent when moving beyond an output, but IMO it should definitely be idempotent for 0 0)

12:04 <ifreund> why won't moving beyond an output be idempotent? The first application should result in the view being send to the edge of the ouput and reduced in size, futher applications should change nothing right?

12:05 <ifreund> anyhow, I agree that moving to 0, 0 should certainly be idempotent. Probably just missing a + border_width on some check

12:05 <bfiedler> if I do set-position -10 -10 it will always crop the window

12:05 <bfiedler> (until at some point the minimum size is reached)

12:06 <ifreund> hmm, ok I see why that's the case but it seems less than ideal

12:07 <bfiedler> Aside from the fact that one shouldn't be doing that I think it's ok

12:07 <bfiedler> we are moving the top-left corner, so if you repeatedly put that somewhere where it will be cropped it's on yourself

12:09 <ifreund> perhaps we just shouldn't accept negative coordinates?

12:09 <bfiedler> That would also be fine with me

12:11 <ifreund> I don't really see a reason to do so, and I think idempotency isn't an issue on the bottom/right edges as after the first operation subsequent operations will do nothing

12:11 <ifreund> so if we fix the border_width issue and forbid negative coordinates set-position should be idempotent I believe

12:12 <bfiedler> agreed. I'll add some input validation to the command then

12:12 <ifreund> cool