A recent article suggests that comparing floating-point numbers for equality is sometimes acceptable, but industry experts argue this view is overly reductive for computational geometry. Because geometric inputs are inherently inexact, exact floating-point comparisons are fundamentally unworkable for boundary representations, requiring geometry kernels to rely on fuzzy tolerances instead. While field-based representations can bypass these floating-point precision issues, developers note they introduce severe new complexities, such as the difficulty of implementing user-requested fillets in CAD software.