Hi!
DIfferent output backends share similar code. This could mean some new feature or bug fix must be copy/pasted (with little adaptation) in all of them or output will not be consistent across them.
What if printing would be done in 2 stages:
1) general vector format like in [vector-]render.c taking care of as much details as possible,
2) rendering vectors (with as few of special handling as possible) in a particular vector format like eps, svg or raster format like png.
If details could not be abstracted in stage 1 to make the stage 2 trivial enough, then it doesn't worth it. In this is the case, maybe you're aware of a better approach?
Thanks,
Andrey
Definately agree - see my response to #98.
Robin.
This is now in place, although it has not been documented yet. You can find the relevant code in vector.c which puts the symbol into a linked list of rectangles, circles, hexagons and text strings.
Closing as duplicate.