From: chrysn <ch...@fs...> - 2011-03-21 08:54:34
|
hello developers, my patches branch at [1] recently got a few bugfixes / features, namely: * d3eb27ff fix for incorrect api usage resulting in non-black background pixels * a3f43143 narrower sampling raster to avoid missing narrow lines * d2a34132, 056395a7 make sure an outline is always milled completely. this rarely shows up with voronoi style outlines as they always mill everything twice, but does so with lower limits (e.g. for large unwired regions on a board). the fixes make sure the outline is completely closed even for small dpi (c5578fca) and the movement isn't smoothed over by the machine. * f4dfbacd configure smoothing in a dpi-dependent way instead of a hard-coded distance; thus, movements are always smooth instead of the aliased zig-zagging. the discrepancies are always below a pixel now. (this was actually a side effect of experimenting with the above) * 55d33410, 829e20be replace the optimizations of f87577f5, which optimize out one point too much. i hope the new version is more readable too. i would have removed the code duplication if it was not explicitly commented to be on purpose: /** if we're cutting, perhaps do it in multiple steps, but do * isolations just once. i know this is partially repetitive, but * this way it's easier to read */ i respect the author's choice, but would kindly ask to reconsider, given the amount of code duplicated is larger now. i'd like to use the opportunity to point out 91b2b2dd as well, which fixes the documentation to reflect the current code. regards chrysn [1] git://gitorious.org/pcb2gcode/pcb2gcode.git -- Reflection. Surprise. Terror. For the future. -- Kosh |