Hello Raster devs,<br><br>I'm setting up regression tests for rotation, skew, etc.<br><br>I'm finding that the values of rotation, skew, and pixel size are off by very small floating point errors (on the order of 3.85e-15). Is it unholy and wrong to do some rounding in the tests, to accommodate these floating point errors in the regression tests?<br>
<br>I'm thinking of something like:<br><br>... where round(st_rotationx(rast)::numeric,15) = 30.0;<br><br>Please advise,<br>David<br>