Jordan Brown

But based on a very quick look at the test results, I have more to do. My guess is that changing the resolution from 100 to 72 subtly changed the...

Yes, that's what I expect. It's not as bad as you might expect, since FreeType thinks it's rendering type that's about 22" tall, so it's dropping from a grid of...

Some thought and research later... I think the 1/64 factor is cancelled on the output side by the coordinates being passed back from FreeType *also* being in 1/64 units. Unfortunately,...

Found the big problem, I think. The csgpngtest tests operate by generating a .csg file and then processing that .csg file. My change didn't fix the CSG dump, so a...

At the moment, all of the "dump" tests fail because they dump the "corrected" value, which does not match the golden reference. Emitting "em" as described above won't fix that,...

The push above changed strategies: instead of correcting the "size" parameter at the top, the top records whether it was set using size= or em= and the bottom uses that...

And I #if'ed out the raw stuff. Still need to decide what the final answer is there, but #if'ing it out should let the fontmetrics() test pass.

Any opinions on what I should do with the "raw" stuff? Rip it out, leave it under #if, or do something more elaborate? Will it be helpful for me to...

That push ripped out the "raw" stuff. It was easy enough to add; if there's a need for it I can add it again.

I agree that underline position and thickness are not debug information. (Or, rather, they would be in their original units, but not if converted to OpenSCAD units.) I can add...