Tom HappVerified account

sketching out an algorithm for uh, figuring out how much to reduce font size by to get an oversized block of text to fit in a text box. its more complicated than you think

if the text is word-wrapped and does not contain any manual line breaks, you *should* be able to measure the vertical overflow to determine how much to reduce the font size by (it'd be proportional to the square of that value I think, there's a square/sqrt in there somewhere)

if there's a few manual line breaks in there, those contribute linearly to the vertical overflow, not proportional to the square, so it gets weirder in that case. I think iterating that above thing a few times would maybe converge?

but I'm not sure that would converge fast enough (figuring out how much overflow there is on a specific font size is not a trivial operation, so if that would take 100 iterations to converge it might actually be noticeable)

but shrinking linearly would 100% get text to fit, so that would be a lower bound on the font size, and shrinking sqrtdly would be an upper bound, could binary search between those (picking whichever option is "closest" each time) and probably converge in ~5-10 iterations

literally just doing a dumb binary search makes it converge in ~10 iterations my other things didn't seem to be good enough estimations