>>Does that seem right to you? I'm confused. Yes, 2" wont fit, but are we measuring the item or the box with rounding?
No it does not. I didn't test thoroughly, and focused mostly on the padded FRE and Reg A, but again, it looks like someone with no "feel" for shipping wrote the code, and either used an incorrect packaging size reference table, or there is a bug as simple as rounding down as you observe. (I also considered for a moment that the algorithm might be using volume instead of dimensions to compare for some unknowable reason, but dismissed that concept.)
Since fractional dims aren't used, the algorithm had to choose rounding up or down, and it looks like they chose "round down" always - which is the antithesis of a shipper's outlook - either the actual size was a bit over 1/2" and I rounded up when entering dimensions, or "I can smash it, stuff it, sit on it, etc and it will fit in that SFRB with only a little (legal) bulge". The rounding down approach prohibits that (without clicking 20 times to switch to USPS packaging tab)
It's for example especially hard to put size limits on a PFRE going by the flat dimensions - some legendary stuffing exploits are possible with those, and I doubt that was considered. (didn't test all possibilities, but it does permit 12x9x4 which is pretty generous)
One mistake this approach makes is assuming the seller put exact, smallest possible pkg dimensions in the listing which is rarely the case - one chooses safe dimensions for the most likely to be used box on hand, which often may be larger than the item, and totally wrong if other options are being considered.
Idiot proofing the label flow is fine as long as those of us that still want a sharp tool we can use to be flexible as we can with the classic label flow, but have an option to turn off the nanny state feature ("yes I agree to take all responsibility if I pick an incorrect package" - make it a settings option the noobs will be unlikely to find. "Bubble wrap" mode and "sharp pointy surgical scalpel" mode)
