How does analogue vs digital (assuming you mean that analogue) matter here? You'd also want to specify which "digital physics" you want to rely on.

We don't need to think along those lines to consider a physically real idealized geometrical object.

Here I'll recast your square as a planar slice through a real object in our (3d+1d) universe.

With reasonable assumptions about the local behaviour of spacetime, you can't construct a physically perfect square because of the small-length-scale behaviour of matter rather than that or the presence of infinities of infinitesimals. The assumption here is that locally around the square the spacetime is Minkowskian, which for a small and light square is practically guaranteed everywhere outside the event horizons of black holes and far from the hot dense phase of the universe. "Practically guaranteed" here means that in spite of plausible theoretical objections to the perfectness, there exists no mechanism to show that the object is not a perfect square (i.e., such objections are conjectural physics leaning hard on a purely mathematical argument).

"Digitalization" and discretization arguments run into observational problems right away.

The wavelength of light is not quantized, for example, and is not clearly bounded in the infrared. (In the ultraviolet you get pair production and gravitational collapse, however.) While one could make an argument that at the emission event of every photon there is not an infinity of available photon wavelengths, you would have to forbid infinitesimal spacetime intervals to avoid emitted photons stretching into arbitrary wavelengths under the metric expansion of the universe, and the only practical way of doing that is through discretization (with or without emergence).

There is a substantial weight of evidence that suggests that if spacetime (or whatever it emerges from) is discretized, the discretization length is not significantly above the Planck length; notably gamma rays, x-rays, light and radio from violent astrophysical events do not appear to win races against each other in an ordered fashion.

So there may be infinitesimal spacetime intervals, and so any quantity that depends on spacetime intervals has an infinite set of values to choose from.

Absent a description of gravity-matter interaction when matter's small-scale behaviours are relevant, we cannot really preclude the possibility of building a perfect square in an unreasonable (but nevertheless physical) local patch of dynamical spacetime, such that the microscopic behaviour of the latter offsets the odd quantum mechanical contributions to the square. At best we can try to place limits on how long such a system could survive (it could last a very long time in the cold sparse phase of the universe).