Problem: For data from very long slit spectrographs, e.g., slit length >> 60", 'global' smile might be too strong for spectroflat to correct. Suggestion: Add a global smile correction as a first step.