kicad-developers team mailing list archive

[jp charras <jp.charras@xxxxxxxxxx>]

Le 17/06/2015 00:11, Chris Pavlina a écrit :
> Hi all,
> 
> While working on that field autoplace feature, I noticed that 
> LIB_PART::GetBodyBoundingBox was returning incorrect bounding boxes when 
> the parts contained a LIB_TEXT. Further investigation revealed that 
> there was an extra negation of the vertical dimension somewhere in the 
> mix. The trouble is, there appear to be multiple of these, because 
> adding or removing negations would fix it in this place, but would cause 
> other troubles (text rendering in the wrong spot in libedit, or clicks 
> not being received). The only way I found that worked was to 
> special-case LIB_TEXT inside LIB_PART::GetBodyBoundingBox, negating both 
> the Y position and Y size. Ugly hack attached.
> 
> Anyone familiar enough to have a guess as to what the "real" problem is, 
> without me spending hours digging into the entirety of the bounding box 
> code? The ugly hack works, but... ew.
> 
> --
> Chris

For historical reasons, the Y axis is bottom to top for coordinates of
symbols (graphic items, texts ..) in libraries, and is top to bottom in
schematic.

When drawing a symbol, a matrix transform is applied to coordinates, to
take in account the Y axis orientation, as well as component orientation
(rotation, mirroring).


-- 
Jean-Pierre CHARRAS